CN120600192A - Long-distance underground culvert structure health assessment method - Google Patents

Abstract

The invention discloses a long-distance underground culvert structure health assessment method which comprises the steps of carrying out fusion processing based on acoustic and optical detection data to identify structural defects, carrying out underwater-dry land conversion and correction of strength values based on underwater rebound and coring test data to determine corrected structural strength and material degradation parameters, carrying out structural evolution modeling and mutation risk identification by combining defect, degradation and history monitoring data to obtain structural evolution features and mutation risks, and carrying out full-field state assessment by integrating the corrected structural strength, evolution features and mutation risks to generate structural health assessment and risk early warning results. The invention can realize accurate, dynamic and full-field evaluation of the health state of the culvert, and improves the reliability and early warning capability of the evaluation result.

Description

Long-distance underground culvert structure health assessment method

Technical Field

The invention belongs to the field of civil engineering and structural health monitoring, and particularly relates to a long-distance underground culvert structural health assessment method.

Background

Long-distance underground culvert structures are usually buried deep underground, cross complex geological and environmental units such as rivers, highways and the like, and are used for a long time under severe conditions of full water, high flow velocity and sand inclusion water flow. The factors such as hydraulic flushing, material aging, geological change and the like which are accumulated in daily life inevitably cause concrete cracking, material degradation, joint water stopping failure, soil body leakage around the structure, piping channels and the like. Once the structure is suddenly damaged, not only can the engineering stop be caused, but also the secondary risk of the surrounding environment can be caused. Therefore, the system, the accuracy and the prospective health state evaluation are carried out on the long-distance underground culvert, the potential risk is found in time and the early warning is carried out, and the system and the method have theoretical significance and engineering application value for guaranteeing the safe and efficient operation in the whole life cycle.

Currently, detection and evaluation technologies for long-distance underground culverts have been developed. In terms of detection means, the conventional manual dive probe approach is gradually being supplemented or replaced by an underwater Robot (ROV) with various sensors mounted thereon. ROVs are typically equipped with high definition optical cameras, forward looking sonar or multi-beam sounding systems, which can provide initial investigation of the condition of silting in a hole, apparent concrete disease (e.g., cracking, spalling) without evacuating the culvert. In terms of structural performance evaluation, in-situ non-destructive testing (NDT) techniques have been applied, such as sampling concrete strength using rebound in areas with operating conditions. Meanwhile, fixed monitoring sensors such as osmometers, strain gauges, displacement meters and the like are arranged at key positions of the culvert so as to acquire long-term monitoring data of key physical quantities of the structure. In addition, in the analysis method, a numerical simulation technology based on a Finite Element Method (FEM) is widely used for analyzing seepage field and stress field distribution of culverts under design load and boundary conditions, and theoretical basis is provided for understanding basic mechanical behaviors of structures. The application of the techniques and methods forms the basis of the current culvert health assessment work.

However, although the detection of culverts is realized to a certain extent in the prior art, the detection method still has significant limitations in the aspects of deep fusion of multi-source heterogeneous data and space-time evolution evaluation of structural health states. Specifically, in the data layer, detection data (such as sound, light and mechanics) from different sources lack of uniform physical models to perform space-time alignment, underwater environment interference correction and defect influence quantification, so that fusion result accuracy is low and physical significance is ambiguous, and information island and evaluation misalignment are formed. In the evaluation framework layer, the existing method usually stops at static and discrete description of the current state, so that the dynamic evolution rule of the structural health along with the multi-field coupling effect cannot be revealed through a unified evolution function, and the sparse monitoring data cannot be reliably reconstructed into the full-field continuous state based on physical constraints. This dual limitation of data and framework makes it difficult in the prior art to accurately and timely pre-warn of the potential risk of long-distance culverts.

Disclosure of Invention

The invention aims to provide a long-distance underground culvert structure health assessment method for solving the problems in the prior art.

The technical scheme is that the method for evaluating the health of the long-distance underground culvert structure comprises the following steps:

Based on the acoustic and optical detection data, fusion processing is carried out, and structural defect data comprising soil body types, defect types and geometric parameters are identified and output;

Based on pre-stored underwater rebound and coring test data, combining structural defect data, performing underwater-dry conversion and correction of strength values, and determining corrected structural strength and material degradation parameters;

combining the structural defect data, the material degradation parameters and the prestored historical monitoring data, carrying out structural evolution modeling and mutation risk identification, and obtaining the evolution characteristics and mutation risks of the structure;

and (3) comprehensively correcting the structural strength, evolution characteristics and mutation risk, and performing full-field state evaluation to generate structural health evaluation and risk early warning results.

The method has the beneficial effects that the accurate, dynamic and full-field evaluation of the health state of the culvert can be realized, and the reliability and the early warning capability of the evaluation result are improved.

Drawings

Fig. 1 is a flowchart illustrating steps of a method for evaluating the health of a long-distance underground culvert structure according to an embodiment of the present application.

Fig. 2 is a flowchart of steps for obtaining evolution characteristics and mutation risk of a structure according to an embodiment of the present application.

Fig. 3 is a flowchart of steps for generating an evolution time sequence according to an embodiment of the present application.

Fig. 4 is a flowchart of a step of dynamically calibrating an evolution weight coefficient according to an embodiment of the present application.

Detailed Description

In order that those skilled in the art will better understand the present invention, a technical solution in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in which it is apparent that the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present invention without making any inventive effort, shall fall within the scope of the present invention.

It should be noted that the terms "comprises" and "comprising," along with any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed or inherent to such process, method, article, or apparatus.

It was found in the study that the data-level islanding and misalignment problems resulted in insufficient accuracy and integrity of the assessment basis. The multisource data obtained by the prior art method is physically split and uncalibrated. The physical basis of the acousto-optic data fusion is weak. The optical and acoustic sensors carried by the ROV have inherent deviation in time and space due to physical separation and platform movement, and the feature fusion in the real sense can not be realized by simple image superposition. More importantly, the attenuation and scattering rules of acoustic and optical signals by materials and water flow conditions of different structural partitions (such as a top arch and a side wall) are completely different, and the spatial heterogeneity cannot be adapted by adopting a fixed fusion strategy, so that the defect identification accuracy is limited. The interference of the underwater environment with the in-situ detection results is not precisely quantified. For example, the underwater rebound method detects the strength of concrete, and the rebound value is simultaneously subjected to the compound influence of multiple physical effects such as water pressure, water temperature, water flow flushing, water medium coupling and the like, and the direct use of a land conversion formula can cause serious misjudgment of the structural strength. Existing methods lack a physical model for decoupling and comprehensive compensation of these interfering factors. The geometric information of the defect is disjointed from its mechanical influence. The correlation between the detected defects such as cracks, flaking and the like, their geometric information such as position, size and the like, and their actual weakening effect on the strength of the surrounding material is ambiguous. During evaluation, only qualitative description or rough reduction can be carried out, and spatial distribution of defect influence cannot be quantitatively and continuously reflected, so that the local risk evaluation is out of alignment.

In addition, static and discrete problems at the evaluation framework level limit the global and prospective nature of the evaluation results. Existing assessment modes tend to be static patches, rather than an organic whole. A unified space-time evolution assessment framework is lacking. Assessment is usually stopped by generating a current defect profile or intensity profile, and cannot answer the core questions of how the health of the structure as a whole evolves from the past to the present. For the processes of multiple physical fields such as seepage, stress, material degradation and the like, the existing method is difficult to unify the processes in an evolution function capable of quantifying accumulated damage, and the dynamic evolution rule of the structural health state cannot be revealed. The evaluation results were spatially discrete and lacking in physical connotation. Whether in situ or sensor monitoring, the data obtained is essentially discrete point or line information. How to scientifically reconstruct the full-field state distribution of the whole culvert from the sparse data is a key problem. The traditional spatial interpolation method only considers geometric distance, ignores the physical rule (such as diffusion and propagation) of the evolution state in the structural medium, so that the reconstructed full-field state diagram can be far away from the actual physical process, and the concentrated region and the propagation path of risks cannot be accurately identified.

As shown in fig. 1, a method for evaluating the health of a long-distance underground culvert structure is provided, which comprises the following steps:

Based on the acoustic and optical detection data, fusion processing is carried out, and structural defect data comprising soil body types, defect types and geometric parameters are identified and output.

In this embodiment, the acoustic detection data includes multi-beam and side-scan echo time difference, amplitude and angle of incidence information, ultrasonic pulse-echo data, acoustic tomography data, and back-scatter or scatter cross-section data for evaluating substrate and defect scatter intensities at different frequencies. The optical detection data comprise visible light high-resolution images and video frames, laser radar point clouds, multispectral or hyperspectral images and infrared thermal imaging data. This step is the starting point for data-driven analysis, aimed at overcoming the limitations of a single data source in complex underwater environments (such as high turbidity bodies of water) by fusing the advantages of different sensors, thereby accurately identifying and locating apparent defects of culverts, such as cracks, spalls, leaks, etc. The structural defect data provides key input information for subsequent intensity correction and evolution analysis.

Based on pre-stored underwater rebound and coring test data, combined with structural defect data, performing underwater-dry conversion and correction of the strength value, and determining corrected structural strength and material degradation parameters.

In this embodiment, this step is intended to obtain the actual performance parameters of the structural material. Considering that the culvert is in an underwater environment for a long time, the conventional dry land detection method is not applicable any more. Therefore, the influence of the underwater environment (such as water pressure and water temperature) on rebound detection can be quantitatively compensated by constructing a physical model, so that the underwater-dry land conversion of the strength is realized. And carrying out local correction on the intensity values around the defects by utilizing the structural defect data to finally obtain an intensity distribution diagram capable of reflecting the real state of the structure, and calculating the material degradation parameters based on the intensity distribution diagram.

And carrying out structural evolution modeling and mutation risk identification by combining the structural defect data, the material degradation parameters and the prestored historical monitoring data to obtain the evolution characteristics and the mutation risk of the structure.

In this embodiment, the purpose of this step is to reveal the dynamic law of the change of the structural health state over time and the potential risk of mutation. Specifically, by integrating multi-source information, an evolution model capable of reflecting the coupling effect of multiple physical fields such as permeation, stress and the like is constructed. By solving the model, the evolution characteristics of the accumulated damage degree of the quantitative structure can be obtained, and the risk transition moment possibly causing structural failure, namely the mutation risk, is identified by carrying out differential analysis on the evolution process.

And (3) comprehensively correcting the structural strength, evolution characteristics and mutation risk, and performing full-field state evaluation to generate structural health evaluation and risk early warning results.

In this embodiment, the evaluation results of the multiple dimensions obtained in the foregoing steps are integrated to form a global and unified health status evaluation. Discrete monitoring point information can be expanded into a continuous state field covering the whole culvert through a sparse data reconstruction technology. On the basis, a nonlinear fusion model is adopted, the structural strength representing the current performance, the evolution characteristic representing the historical evolution and the mutation risk representing the future risk are comprehensively calculated, and finally, the evaluation result is intuitively presented in the form of a comprehensive health index distribution diagram and the like, and a grading early warning scheme is provided.

According to one aspect of the application, before the fusion processing is performed to identify and output the structural defect data, the method further comprises the step of dynamic time-space alignment of the acousto-optic data, so as to solve the difference between the acoustic sensor and the optical sensor in the aspects of physical position, propagation medium speed, dynamic change of a motion platform and the like, and enable the data base of the subsequent fusion processing to be consistent in time-space. Specifically, the alignment step includes:

based on the propagation velocity difference of the acousto-optic light in the aqueous medium and the spatial distance between the sensors, a base propagation time difference is calculated.

In this embodiment, the spatial distance d between the sensors may be obtained by reading a sensor layout file of an underwater Robot (ROV) system, specifically, the euclidean distance d=sqrt[(x_s-x_c)²+(y_s-y_c)²+(z_s-z_c)²]. between the central coordinate (x _s,y_s,z_s) of the acoustic sensor and the central coordinate (x _c,y_c,z_c) of the optical camera, where the sound velocity v _s and the light velocity v _c in the aqueous medium may be calculated according to environmental parameters such as water temperature and water depth monitored in real time through an empirical formula. For example, the sound velocity v _s can be calculated from v _s=1449.2+4.6T-0.055T²+0.00029T³ + (1.34-0.01T) (S-35) +0.016h, where T is water temperature, S is salinity, and h is water depth. The base propagation time difference Δt _base is Δt _base=d(1/v_s-1/v_c).

And accounting at least one disturbance effect introduced by the dynamic change of the detection equipment, and obtaining a dynamic time difference correction term.

In the present embodiment, the disturbance effect preferably includes a movement velocity effect of the underwater robot or a beam spread effect of the acoustic sensor. For example, the correction term Δt _motion introduced by the motion velocity effect can be denoted Δt _motion=d·v_ROV/(v_s·v_c), where v _ROV is the real-time velocity vector of the ROV. The correction term Δt _spread introduced by the beam spread effect can be expressed as Δt _spread=d·tan(θ_s/2)/v_s, where θ _s is the beam angle of the acoustic sensor.

And superposing the basic propagation time difference and the dynamic time difference correction term to generate the comprehensive time difference compensation quantity delta t _total. For example, Δt _total=Δt_base+Δt_motion+Δt_spread.

And carrying out translation alignment on the acquisition time axis of the acoustic data and the optical detection data according to the comprehensive time difference compensation quantity.

Specifically, the time stamp of the acoustic data may be shifted by t _{s_new}=t_{s_old}+Δt_total and the shifted section overlapping in time with the optical data may be resampled to unify the time resolution. Wherein t _{s_old} is a time stamp of the original acquisition of the acoustic data, and t _{s_new} is a new time stamp of the acoustic data after the comprehensive time difference compensation is completed on the acoustic data time axis.

According to the embodiment, the dynamic time difference compensation quantity is calculated by comprehensively considering the acousto-optic propagation speed difference, the ROV motion speed vector and the acoustic wave beam diffusion effect, so that the accurate time-space alignment of multi-sensor data on the mobile detection platform is realized. The problem of defect positioning deviation of the ROV caused by acoustic data lag is solved. After dynamic compensation, the defect positioning precision is improved, so that parallel cracks can be accurately distinguished, and reliable space coordinate information is provided for subsequent accurate repair.

According to one aspect of the application, after space-time alignment is completed, fusion processing is performed to identify and output structural defect data, specifically including:

and respectively analyzing the acoustic characteristic spectrum and the visual characteristic spectrum from the acoustic detection data and the optical detection data.

For example, the time aligned acoustic data is subjected to short-time Fourier transform, frequency domain features are extracted to form an acoustic feature map, and the time aligned optical data is subjected to edge detection and texture analysis to obtain a visual feature map.

According to the structural drawing of the culvert, the culvert body is spatially divided into a plurality of structural partitions with different physical properties.

This division is to achieve adaptive adjustment of the subsequent fusion parameters, as the material and water flow environments of the different partitions have different effects on the acoustic and optical signals. For example, the culvert cross-section may be divided into a top region, a sidewall region, and a bottom region based on structural mechanics characteristics.

For each structural partition, the partition specific fusion parameters are generated in a self-adaptive operation mode according to the specific materials and the water flow environment of the structural partition.

The embodiment is a key for realizing intelligent fusion. The method comprises the step of extracting partition state parameters representing the physical essence of each structure partition from design data and environment monitoring data corresponding to the structure partition. The set of parameters includes at least two of the material properties represented by the concrete thickness or rebar distribution of the zone and the hydrodynamic characteristics represented by the water flow velocity or Reynolds number of the zone. Substituting the partition state parameters into a preset physical model for resolving, and generating unique partition specific fusion parameters for the partition, wherein the unique partition specific fusion parameters are used for respectively weighting the acoustic feature spectrum and the visual feature spectrum. For example, an acoustic attenuation coefficient model α _i=α₀×exp(-β_a×h_i)×(1+0.1×ρ_i)×(1+0.05×Re_i) and an optical scattering coefficient model β _i=β₀×(1+γ_s×TSS_i)×(1-0.2×h_i/h_max) may be established) and normalized, where h _i is the concrete thickness, h _max is the concrete maximum thickness, ρ _i is the reinforcement ratio, re _i is the reynolds number, TSS _i is the suspended matter concentration, α ₀,β_a,β₀,γ_s is the model coefficient, and so on.

And in each structural partition, weighting and fusing the acoustic characteristic spectrum and the visual characteristic spectrum by using partition specific fusion parameters corresponding to the structural partition to generate a fusion characteristic spectrum, identifying defects based on the fusion characteristic spectrum, and finally outputting structural defect data. The fusion formula may be expressed as F _i=α_i×A_i+β_i×V_i, where a _i and V _i are normalized acoustic and visual features, respectively.

According to the embodiment, the acousto-optic fusion weight is calculated in an adaptive mode according to physical parameters such as concrete thickness, reinforcement distribution, water flow Reynolds number and the like of each partition, and the problem that detection conditions of different parts of a culvert are huge in difference is solved. The optical weight is increased in the top low flow rate region to obtain a clear crack image, the acoustic weight is increased in the bottom high sand-containing water flow region to penetrate the turbid medium, and the equilibrium fusion strategy is adopted in the side wall transition region. The whole-course defect detection rate of the long culvert is improved, and reliable data support is provided for preventive maintenance.

Further, after the weighted fusion, the method further comprises the steps of verifying and correcting the fusion result so as to improve the robustness of the fusion result. The method comprises the following steps:

And aiming at the specific structural partition, calculating cross-modal mutual information between the acoustic feature spectrum and the visual feature spectrum corresponding to the specific structural partition, and taking the cross-modal mutual information as a quantization index for measuring the consistency of the two modal data. The greater the mutual information MI (A _i,V_i), the greater the consistency of the information provided by the two data sources. Optionally, MI (a _i,V_i) = Σp (a, v) log [ p (a, v)/(p (a) p (v)) ], where p (a, v) is a joint probability that the acoustic feature value falls at the quantization level a and the visual feature value falls at the quantization level v, p (a) is an edge probability that the acoustic feature falls at the quantization level a, and p (v) is an edge probability that the visual feature falls at the quantization level v.

When the cross-modal mutual information is lower than a preset confidence threshold (for example, MI < 0.3), judging that modal information conflict exists in the partition. Typically meaning that one of the sensors is severely disturbed or disabled in this area.

For the subarea with modal information conflict, the fusion strategy is automatically switched into a preferred selection strategy by weighted fusion, namely, the party with higher confidence (such as the party with better signal quality or characteristic strength) is selected from the acoustic characteristics and the visual characteristics, and the party with higher confidence is used as the final fusion result of the subarea to correct the fusion characteristic map. The method can effectively avoid pollution to the final fusion result due to poor quality of single-mode data, and improves the accuracy of defect identification.

The embodiment ensures that the fusion result has reliability. When the water flow disturbance at the culvert curve causes the optical image blurring but the acoustic signal is clear, the system intelligently selects acoustic data, and the complementary advantages of the two modes are fully utilized in the straight-line water flow stable region. The false alarm rate of fusion detection is reduced, and meanwhile, high sensitivity is maintained, so that the method is particularly suitable for complex environments with frequent changes of detection conditions in long-distance culverts.

According to one aspect of the application, the step of performing the underwater-dry conversion and correction process of intensity values comprises:

constructing and calibrating an underwater-dry land comprehensive conversion coefficient based on coring test data;

The underwater-dry land comprehensive conversion coefficient in the present embodiment aims to compensate for the influence of the underwater environment, so as to systematically eliminate the interference of the underwater environment on the rebound detection result, and specifically includes:

And respectively quantifying and determining independent environment correction factors aiming at physical effects of the underwater environment. The set of environmental correction factors includes at least two, and preferably all four of the following:

A pressure correction coefficient k _p characterizing the effect of hydrostatic pressure on rebound. Since the hydrostatic pressure p _w will generate a pre-stress on the concrete surface, increasing the energy dissipation of the rebound impact, a correction is required. Where p _w=ρ_w×g×h,ρ_w is the density of water (e.g., 1000kg/m ³), g is the gravitational acceleration (about 9.8m/s ²), and h is the detection point water depth. The pressure correction coefficient may be calculated by model k _p=1/(1+λ_p×p_w), where λ _p is the experimentally calibrated pressure influence coefficient (e.g., λ _p=2×10^-6/Pa).

And a temperature correction coefficient k _T for representing the influence of water temperature on the elastic modulus of the concrete. The water temperature T affects the modulus of elasticity of the concrete and thus the rebound value. The correction factor may be calculated from k _T=[1+α_T×(T-20)]×k_age, where a _T is the temperature sensitivity factor for the elastic modulus (e.g., a _T=-0.0004/℃）,k_age is an age correction term that accounts for the effects of early hydration reactions, e.g., k _age=0.9+0.1×(t_age/90 when the concrete age t _age is less than 90 days).

And a scouring correction coefficient k _e for representing the influence of water flow scouring on the surface hardness of the concrete. The long-term water flow scouring can abrade the surface layer of the concrete, and the hardness of the concrete is reduced. The effect can be quantified by a model k _e=1-β_e×η_e, where η _e is the rate of loss of surface hardness, estimated from the ratio of cumulative flush depth Δ _e to aggregate average particle size d _agg (η _e=Δ_e/d_agg), Δ _e in turn being related to water flow velocity v and length of service t _service, and β _e is the flush influencing factor (e.g., β _e =0.15).

The medium coupling coefficient k _c, which characterizes the influence of the aqueous medium on the rebound impact energy transfer. The presence of water increases damping during rebound impact and reduces energy transfer efficiency. The effect can be corrected by k _c, for example k _c can be calculated from the baseline efficiency ratio and taking into account the water depth effect, such as k _c =1.31× (1-0.001×h).

And generating the underwater-dry land comprehensive conversion coefficient by coupling and integrating all the environmental correction factors.

Specifically, the integration step includes performing product operation on a plurality of independent environment correction factors in the group to obtain a preliminary comprehensive coefficient k=k _p×k_T×k_e×k_c. And further introducing a second-order interaction correction term delta k for representing coupling influence among different physical effects, and finally correcting the preliminary comprehensive coefficient to generate an underwater-dry comprehensive conversion coefficient k _final. Because the physical effects are not completely independent, for example, the coupling effect of high water pressure and low temperature may be greater than the product of the independent effects of both. The second order interaction correction term may be expressed as ak=0.05× (k _p-1)×(k_T -1), and the final composite conversion coefficient is k _final =k× (1+ak).

And converting the underwater rebound into a preliminary dry land equivalent intensity value by using an underwater-dry land comprehensive conversion coefficient.

For example, the underwater rebound original value R _w can be converted to a preliminary dry equivalent intensity value f '_c using the formula f' _c=k_final×(A×R_w -B), where A and B are calibration parameters for the resiliometer.

And carrying out local correction on the preliminary dry equivalent strength value according to the defect type and the geometric parameter to obtain corrected structural strength, and generating a material degradation parameter based on the comparison of the corrected structural strength and preset design strength.

In this embodiment, the step of locally correcting specifically includes:

For each defect in the structural defect data, constructing a three-dimensional influence field model I (r) with the influence intensity attenuated along with the increase of the space distance according to the geometric parameters of the structural defect data. For example, the model may be expressed as a gaussian decay function I (r) =i ₀×exp(-r²/2σ², where r is the distance from any point in space to the center of the defect, I ₀ is the intensity of influence of the center of the defect (normalized to 1), σ is the radius of influence, and its value is related to the length l and width w of the defect, e.g., σ=sqrt (l ²+w²)/4.

Based on the defect type, setting a basic strength reduction coefficient representing the maximum influence degree of the defect type. For example, for a crack, its reduction coefficient k _crack may be related to the crack width w, k _crack=1-0.15×(w/w_cr)^0.5, where w _cr is the critical width, and for spalling, its reduction coefficient k _spall may be related to the spalling depth d.

And fusing the three-dimensional influence field model and the basic intensity reduction coefficient, and calculating a defect correction coefficient field k _d (x, y, z) which continuously changes in space. Specifically, for a single defect, k _d(x,y,z)=1-(1-k_type)×I(r)/I₀, where k _type is the base intensity reduction coefficient for that defect type. Optionally, when the three-dimensional influence field models of the plurality of defects overlap in space, the method further comprises the steps of carrying out probability superposition operation on defect correction coefficients corresponding to the independent defects in a space overlapping region, generating a final correction coefficient k _total capable of reflecting the coupling influence of the plurality of defects, and correcting intensity values in the overlapping region by adopting the final correction coefficient. The probability superposition may be denoted as k _total=1-Π(1-k_i), where k _i is the correction factor caused by the ith independent defect at that point.

And applying the defect correction coefficient field to the preliminary dry equivalent intensity value to obtain the corrected structural intensity. And the structural strength f _c=k_d×f'_c is used for finishing the strength correction of the defect influence area.

The present embodiment establishes a high-precision underwater-dry strength conversion model. The model considers the rebound value reduction caused by every 10 meters of water depth increase, the elastic modulus change when the water temperature deviates from 20 ℃, the surface hardness attenuation caused by long-term water flow scouring and the influence of water serving as a coupling medium on the impact energy transmission efficiency, and reduces the error of underwater concrete strength evaluation. The long-distance culvert can obtain the whole-course accurate intensity distribution without taking a large amount of cores, the detection efficiency is improved, and the cost is reduced. The method realizes the fine quantification of the influence of the defects on the concrete strength by establishing a three-dimensional Gaussian attenuation field model influenced by the defects and adopting probability superposition to process a multi-defect overlapping area. And the elliptical influence area, the peeled hemispherical influence area and the cooperative degradation effect of multiple defects of the crack along the trend are considered, so that compared with the traditional two-dimensional fixed range, the spatial resolution of strength evaluation is improved. The composite influence of the underwater environment and the local defects on the strength detection is systematically eliminated, a more accurate and reliable structural strength evaluation result is obtained compared with the traditional method, and a solid data base is provided for subsequent structural health evaluation.

As shown in fig. 2, according to an aspect of the present application, the steps of performing structural evolution modeling and mutation risk identification, and obtaining the evolution features and mutation risk of the structure include:

and integrating the structural defect data, the material degradation parameters and the historical monitoring data, constructing and solving an evolution function of multi-physical field coupling, and generating an evolution degree time sequence of the accumulated evolution degree of the quantized structure.

In this embodiment, the evolution degree time sequence Φ (t) is a dimensionless parameter with a value range between [0,1], wherein 0 represents the structural integrity state, 1 represents the structural complete failure or reaches the evolution end stage, and the value of the evolution degree time sequence Φ (t) intuitively reflects the accumulated damage or health degradation degree of the structure.

Further, as shown in fig. 3, the steps of constructing and solving the evolution function of the multi-physical field coupling to generate an evolution time sequence specifically include:

And extracting a plurality of physical field indexes at least comprising the hydraulic gradient i, the seepage flow q and the soil strain epsilon from the historical monitoring data, and counting the time increment sequences of the physical field indexes.

In particular, the historical monitoring data may originate from various types of sensors deployed on the culvert, such as pressure gauges, flow meters, strain gauges, and the like. For evolution analysis, the original time series data are preprocessed, for example, the time resolution is unified to 1 hour, the variation of the original time series data in a time step delta t is calculated, so that a hydraulic slope increment sequence delta i (t) =i (t) -i (t-delta t) is obtained, i (t) is hydraulic slope at time t, delta t is a time step between two adjacent monitoring data, a seepage rate change sequence delta q (t) = [ q (t) -q (t-delta t) ]/A _seep is obtained, q (t) is seepage rate at time t, A _seep is seepage area, and a strain rate increment sequence delta epsilon (t) = epsilon (t) -epsilon (t-delta t), wherein epsilon (t) is soil strain rate at time t, and epsilon (t) = d epsilon (t)/dt.

And dynamically calibrating evolution weight coefficients corresponding to the physical field indexes according to the soil body types and the material degradation parameters.

In this embodiment, the degree of contribution of different physical field processes to the structural degradation is not constant, but is closely related to the specific environment (such as soil type) and the state of the structure (such as material degradation degree). Optionally, as shown in fig. 4, dynamically calibrating the evolution weight coefficient includes configuring a basic weight coefficient for each physical field index according to a soil type, extracting a structural degradation degree index from a material degradation parameter, correcting at least one coefficient value in the basic weight coefficient according to the structural degradation degree index to obtain an adjusted weight, and performing normalization calculation on the adjusted weight to generate the evolution weight coefficient. Specifically, a basic weight coefficient is configured for each physical field index according to the soil type contained in the structural defect data. For example, depending on the soil mechanics characteristics, higher base weights may be configured for hydraulic dip i and seepage q (e.g., α=0.5, β=0.3) for sand more susceptible to seepage damage, and higher base weights may be configured for soil strain ε (e.g., γ=0.3) for clay more susceptible to deformation. A structural degradation degree index (e.g., strength loss rate eta) is extracted from the material degradation parameters, and at least one coefficient value of the basic weight coefficients is corrected according to the structural degradation degree index to obtain an adjusted weight. For example, when the strength loss rate η exceeds a certain threshold (e.g., η > 0.2), it indicates that the structural load capacity is decreasing, which is sensitive to the corresponding change, and the weight of the strain term may be increased at this time, e.g., the corrected weight γ' =γ× (1+0.5 η). And performing normalization calculation on the adjusted weights to enable the sum of coefficient values to be one, and generating final evolution weight coefficients alpha ', beta ', gamma '.

And carrying out weighted fusion on the time increment sequence by adopting an evolution weight coefficient to obtain an evolution increment sequence dphi/dt reflecting the instantaneous evolution rate of the structure.

In the present embodiment, the instantaneous evolution rate can be expressed as dΦ/dt=α ' ×Δi (t) +β ' ×Δq (t) +γ ' ×Δε (t). Alternatively, to eliminate the influence of measurement noise on the calculation result, a kalman filter may be applied to the evolution increment sequence after calculation.

And carrying out numerical integration on the evolution increment sequence along the time dimension to generate an evolution degree time sequence.

Specifically, a numerical integration method such as a trapezoidal integration method may be adopted, and starting from an initial state Φ ₀ (for example Φ ₀ =0), dΦ/dt is accumulated step by step, that is, an evolution time sequence Φ (t) =Φ ₀ +.k (dΦ/dt) dt. Through the step, a curve capable of continuously reflecting the evolution process of the culvert structure from the past to the current health state can be obtained, and a foundation is provided for subsequent evolution stage division and mutation risk identification.

According to the embodiment, the evolution weight coefficient is dynamically adjusted according to the soil type (clay/silt/sand) and the material degradation degree, so that the accurate matching of the multi-physical field coupling model and the actual geological conditions is realized. When the material degradation rate is detected to be more than 20%, the system automatically increases the strain weight, and more attention is paid to the influence of structural deformation, and in a sandy soil area, the hydraulic gradient weight is increased, so that the risk of seepage damage is monitored. The same evolution model can adapt to complex and changeable geological conditions along the culvert, and compared with a fixed weight method, the prediction accuracy is improved, and particularly the evolution evaluation accuracy of key parts such as stratum interfaces is improved.

And carrying out differential analysis on the evolution degree time sequence, and identifying potential mutation moments representing risk transitions.

The aim of this embodiment is to capture, from a continuous evolution curve, nonlinear, non-stationary key nodes that are predictive of the qualitative change of the structural state. Specifically, identifying potential mutation moments that characterize risk transitions includes:

The method can be used for solving the first derivative sequence psiphi/psit and the second derivative sequence psi ²Φ/Ψt² of the evolution degree time sequence phi (t), and respectively representing the evolution rate and the evolution acceleration of the structure, wherein psij is a partial derivative.

In this embodiment, in order to improve the calculation accuracy and avoid the amplification of the numerical error, it is preferable to conduct derivation by a five-point center difference method. For example, the first derivative may be calculated from [ - Φ (t+2Δt) +8Φ (t+Δt) -8Φ (t- Δt) +Φ (t-2Δt) ]/(12Δt). In addition, to eliminate the influence of the noise of the original data on the derivative calculation, a salvinsky-Golay (Savitzky-Golay) filter smoothing process may be performed on the derivative sequence.

And constructing a mutation sensitivity time sequence by calculating the ratio of the absolute value of the corresponding numerical value of the second derivative sequence to the absolute value of the corresponding numerical value of the first derivative sequence at each moment.

Specifically, the mutation sensitivity time sequence S _m(t)=|Ψ²Φ/Ψt² |/|ψΦ/ψt|. The mutation sensitivity is a dimensionless index, the change of the evolution acceleration relative to the current evolution rate is amplified, and the degree of inflection point or abrupt change of the evolution trend is represented in a physical sense. The ratio is small when the structure is in a steady evolution stage, and the evolution acceleration changes sharply when the structure state is about to or is undergoing abrupt change, resulting in a peak value of the ratio.

And comparing each sensitivity value in the mutation sensitivity time sequence with a preset threshold value, and identifying the moment exceeding the threshold value as the potential mutation moment.

In particular, the threshold may be a fixed value (e.g., S _cr =2.5), or in a preferred embodiment, an adaptive threshold is employed. For example, by calculating the moving average value s× _m and standard deviation σ _s of the mutation sensitivity time sequence S _m (t) in a sliding time window, the mutation threshold is dynamically adjusted to S _cr=S*_m+n×σ_s (where n may be 2.5), so that the threshold can adapt to baseline fluctuations in different evolution stages, and the accuracy of identification is improved.

According to the embodiment, the mutation sensitivity index is constructed by calculating the ratio of the second derivative and the first derivative of the evolution degree time sequence, so that the nonlinear mutation points in the culvert permeation evolution process are accurately captured. The index can have a peak value when the evolution rate is changed drastically, and the critical moment of transition from the water seepage stage to the osmotic deformation stage and from the osmotic damage to the channel expansion stage is effectively identified. Compared with the traditional method which only depends on seepage flow or displacement monitoring, the system instability can be predicted when the physical quantity is not changed obviously, the early warning time of major accidents such as culvert collapse is advanced, and the safety guarantee capability of underground water conservancy facilities is improved.

Further, to verify the authenticity of the identified potential mutation moment and explore its physical cause, the method further comprises:

For each identified potential mutation moment, extracting acoustic data segments within a predetermined time window before and after the potential mutation moment from the acoustic detection data. For example, if a potential mutation is identified at time t _m, acoustic waveform data within the [ t _m-300s,t_m +300s ] interval is extracted.

And carrying out frequency spectrum analysis on the acoustic data segment, and decoding whether an acoustic abnormal mode associated with a preset specific failure mechanism exists or not.

The present embodiment is directed to cross-validation of abrupt changes inferred from macroscopic physical field data using acoustic emission signals directly related to physical processes such as structural deformation, cracking, etc. Optionally, the identified acoustic anomaly pattern includes at least one of the following characteristics that, in the high frequency band, a sudden increase in spectral energy occurs with an amplitude exceeding a predetermined decibel value. For example, in the frequency range above 20kHz, sudden increases of more than 10dB above background noise occur, and this mode is usually associated with brittle failure events such as micro-crack propagation of the material, water stop material tearing, etc. In the low frequency range, a sustained jump in vibration intensity significantly above the background noise level occurs. For example, in the frequency band below 100Hz, the vibration energy continues to be greater than 5 times higher than the normal level, and this mode may correspond to an abnormal change in resonance of the structure as a whole, particle washout, or pulsating pressure of the water flow.

If the acoustic abnormal mode exists, the potential mutation moment is updated to be verified mutation risks, and the mutation risks are classified according to the acoustic abnormal mode.

For example, if a sudden increase in high frequency energy is identified, the risk cause may be initially classified as a risk of structural cracking, and if a sudden increase in low frequency vibration is identified, the risk may be classified as a risk of structural instability or washout. The reliability of risk early warning is improved, and a direct physical basis is provided for subsequent treatment decisions.

In the embodiment, the time sequence association verification is carried out on the potential mutation moment calculated by the mathematical mutation theory and the abnormal mode of the acoustic spectrum before and after the moment, so that a double confirmation mechanism of mutation risk is realized. When the second derivative of the evolution degree function has abnormal peaks, the system automatically extracts acoustic data in a corresponding time window to perform spectrum analysis, and identifies high-frequency noise sudden increase (> 20kHz, amplitude increase >10 dB) generated by tearing of the water stop or low-frequency resonance (< 100 Hz) caused by structural vibration, so that abstract mathematical early warning is converted into specific physical damage types. The false alarm rate of mutation in culvert detection is reduced, and early signs of infiltration damage can be identified in advance, so that accurate time window and failure mode information are provided for maintenance decision.

Deducing and generating an evolution rate distribution diagram according to the evolution degree time sequence and the spatial position of the structural defect data;

In this embodiment, the evolution rate of the time dimension is correlated with the spatial position of the structural defect, so as to generate a spatial dynamic evolution field, and visually present the degradation trend differences of different regions of the structure. Specifically, for each defect point x _i, its corresponding local evolution rate Φ (x _i, t) is extracted (i.e., the value of the first derivative ψΦ/ψt at position x _i), discrete Φ (x _i, t) is mapped to the continuous domain Φ (x, t) = Σ _i=1 ⁿλ_iΦ*(x_i, t), where the weight λ _i is determined by spatial covariance function optimization, ensuring that the interpolation results conform to the spatial correlation of the structural material. The method comprises the steps of outputting a space-time evolution rate field phi (x, t) (three-dimensional space plus time tensor), selecting a key time slice t _k (such as a current moment and a historical mutation point), extracting the slice field phi (x, t _k), and generating an evolution rate distribution map through visualization of a Contour filling map (ContourPlot) or a thermodynamic diagram (Heatmap).

And forming the evolution characteristics and mutation risks of the structure together by the evolution degree time sequence, the potential mutation moment and the evolution rate distribution diagram.

In the embodiment, the time sequence evolution, the abrupt node and the space velocity field are fused to form a multidimensional risk criterion, and the support structure is subjected to security grading decision. Specifically, an evolution characteristic report is generated based on an evolution degree time sequence, potential mutation moments and an evolution rate distribution diagram, wherein the evolution characteristic report comprises accumulated evolution phi (t _now) which represents the overall degradation degree of a structure, current rate max phi (x, t _now) which represents the evolution speed of the most dangerous area and mutation risk S _m(t_now) (sensitivity) which represents the probability of the mutation of the approaching state. A dynamic risk classification model is constructed, and a risk level R is defined as a weighting function, wherein the R=w ₁·Φ(t)/Φ_crit+w₂·maxΦ*/Φ*_crit+w₃·S_m(t)/S_cr, the weight distribution (example) is that w ₁=0.3, w₂=0.4, w₃=0.3;Φ_crit is the allowable maximum evolution degree (such as 0.5% of a concrete expansion limit), phi is critical rate (such as rock creep rate >1 mm/day), and S _cr is a sensitivity threshold.

According to one aspect of the application, since the monitoring points are always sparse in the actual engineering, the health state of each point of the structure cannot be directly obtained, and the simple spatial interpolation method ignores the physical rule of state propagation in the structural medium. Thus, full field state evaluation includes a full field reconstruction stage to generate a continuous state distribution. Specifically, the full-field reconstruction phase includes:

and arranging evolution parameters contained in the evolution features and the mutation risks as sparse state monitoring points in the three-dimensional space of the culvert.

In this embodiment, the evolution parameters refer to the evolution degree Φ, the evolution rate ψΦ/ψt, and the like of each key position. The choice of these key locations may be further optimized, for example, by preferentially selecting the region where the evolution rate gradient changes drastically or where the risk of mutation has been identified.

And constructing a physical constraint equation for describing the space-time evolution of the propagation rule of the evolution state in the structural medium.

The present embodiment incorporates engineering experience and physical mechanisms into the spatial interpolation process. In a specific embodiment, the space-time evolution physical constraint equation is a partial differential equation, and the equation is composed of at least a time derivative term for describing the change of the evolution state with time and a space derivative term for describing the propagation and diffusion of the evolution state in space. For example, an evolution propagation equation of the form ²Φ-(1/c²)Ψ²Φ/Ψt² =s (x, y, z) can be constructed, where Φ is the evolution of the spatial position (x, y, z) at time t, which is the unknown to be solved, v ² is the laplace operator, which represents the spatial second derivative term (ψ ²/Ψx²+Ψ²/Ψy²+Ψ²/Ψz²) describing the spatial diffusion or propagation trend of the evolution state, c is the evolution propagation speed, which is a physical parameter, which represents the propagation speed of the degrading or damaging effect in the structural medium, whose value can be calibrated according to different medium types (e.g. sand, silt, concrete), e.g. 2 m/day for the sandy region c, 0.5 m/day for the clay region, t ²Φ/Ψt² is the time second derivative term, which describes the inertia of the evolution state over time, S (x, y, z) is the source term, which represents the endogenous growth rate of the evolution state at the spatial position (x, y, z), e.g. the evolution rate of the evolution state at the point itself may be determined by the degree Φ _local, S (x, y, κ) = _local×(1-Φ_local).

And taking the data of the sparse state monitoring points as a solving boundary or an initial condition, carrying out numerical solving on a physical constraint equation, and generating a full-field evolution degree distribution map which continuously covers the whole culvert.

In particular, the partial differential equation may be solved using a finite element method or a finite difference method. The culvert is subjected to three-dimensional mesh dissection, and preferably, mesh self-adaptive encryption can be performed in a region with a large evolution gradient. The evolution values of the sparse monitoring points are applied to the model as known conditions (e.g., diels boundary conditions). Through iterative solution, the evolution degree phi (x, y, z, t) of all grid nodes in the whole solution domain (namely the whole culvert) can be obtained, and a time-space continuous full-field evolution degree distribution diagram is formed.

In the embodiment, the physical constraint reconstruction from sparse monitoring points to full-field continuous distribution is realized by establishing an evolution propagation partial differential equation comprising a time derivative term and a space derivative term. And determining evolution propagation speed according to the permeability characteristics of different soil bodies, so that the reconstruction result is matched with the measured value at the monitoring point, and the physical rule of permeability damage is also followed in the whole space domain. Compared with the traditional pure statistical methods such as Kerling interpolation, the physical information constraint enables high-risk areas with severe evolution gradient change to be accurately identified under sparse arrangement, spatial resolution is improved, and non-physical oscillation phenomenon possibly generated by statistical interpolation is avoided. The sparse and discrete monitoring data are reasonably interpolated and extrapolated to the whole structural space through a partial differential equation containing a physical mechanism, and the obtained full-field state distribution diagram can reflect the space continuity and evolution rule of the structural health state more truly than the traditional interpolation method.

Further, the evaluation information from different dimensions (structural strength, evolution history and mutation risk) is scientifically fused into a single and visual comprehensive health index, so that the final evaluation of the overall health condition of the structure is realized. Specifically, the full-field state assessment also includes a comprehensive index fusion stage. The fusion phase includes:

And carrying out normalization processing on the full-field evolution degree distribution map, the corrected structural strength, the evolution characteristics and mutation risk indexes in the mutation risk to form a standardized multi-source data set.

The present embodiment is a data preparation work before data fusion, and aims to eliminate dimension and numerical range differences between different data sources. Specifically, all data (e.g., full field evolution profile Φ (x, y, z, t), corrected intensity profile f _c (x, y, z), and mutation risk index P _risk) may be resampled onto a unified three-dimensional spatial grid. Each data source is normalized, e.g., converting the intensity value f _c to a normalized intensity f _norm=(f_c-f_min)/(f_max-f_min ranging between [0,1], where f _min is the minimum intensity value and f _max is the maximum intensity value. The evolution degree and mutation risk index itself has a value range between [0,1], and can be directly used. The normalized dataset {1- Φ _norm,f_norm,1-P_risk } is finally formed, where Φ _norm is the normalized full field evolution, 1- Φ _norm and 1-P _risk represent the health contribution, respectively.

Based on the spatial information entropy of each data source in the standardized multi-source data set, the data fusion weight is dynamically calculated.

This embodiment aims at objectively, data-driven determination of the importance of each evaluation dimension. In particular, the spatial information entropy H _i=-Σp_ij×log(p_ij of each data source i may be calculated), where p _ij is the probability that the value of the ith data source falls within the jth interval. The larger the information entropy, the more abundant the distribution variation of the data source over the whole structural space, the more information content is contained and therefore should be given a higher weight. The initial weight may be set to w _i ⁰=H_i/ΣH_k, where H _k is the spatial information entropy of the kth data source. Optionally, the initial weights may also be modified based on the reliability of the data sources (e.g., reconstruction uncertainty, detection signal-to-noise ratio, etc.), resulting in a final dynamic weight vector [ w ₁,w₂,w₃ ].

And calculating the standardized multisource data set by using the data fusion weight and a nonlinear fusion model containing interactive items so as to generate a comprehensive health index distribution diagram, wherein the diagram forms a core part of the structural health assessment and risk early warning result.

In the present embodiment, the nonlinear model is adopted because there is often a coupling amplification effect between different risk factors, and a simple linear weighting cannot capture such a relationship. The nonlinear model may be represented as h=h _linear+H_interact, where the linear portion H _linear=w₁×(1-Φ_norm)+w₂×f_norm+w₃×(1-P_risk). The interaction term section H _interact then contains product terms of different factors, such as H _interact=λ₁₂×(1-Φ_norm)×f_norm + & gt, where lambda ₁₂ is the interaction coefficient of the i-th and j-th factors.

Further, the nonlinear fusion model further includes a risk avoidance criterion, execution of the criterion including:

Each data source value in the normalized multi-source dataset is checked point by point to determine if any value is below a preset risk threshold. For example, a risk threshold of 0.4 may be set.

If it is determined that the existing value is lower than the risk threshold (for example, the normalized intensity f _norm at a point is only 0.3), the weighted fusion calculation at the point is stopped, and the lowest value at the point is directly adopted as the final comprehensive health index, so that the risk amplification of the short-circuit effect is realized.

That is, in this case, the integrated health index H at this point is directly determined to be 0.3 even if the evolution and mutation risk indexes thereof are high. The risk avoidance criterion follows the barrel theory, so that serious defects in any single dimension are not covered by good performances in other dimensions, and the evaluation result is more conservative and safer. The finally generated comprehensive health index distribution diagram H (x, y, z) has the value range between [0,1], can be subjected to three-dimensional visual display after color coding, and carries out health grading and risk early warning according to different index intervals (for example, H is more than or equal to 0.85 for health, H is more than or equal to 0.70 for basic health, H is less than or equal to 0.85 for basic health, and the like).

The embodiment realizes intelligent fusion and risk amplification of multi-source heterogeneous data. Data sources with high information entropy (such as evolution degree which changes severely in a penetration damage area) automatically obtain higher weight, and when a critical index presents a dangerous signal, the system actively adopts the most conservative evaluation strategy. The subjective problem of traditional expert weighting method is solved, and meanwhile, local serious degradation can not be covered up due to normal other indexes through the risk avoidance criterion, so that the reliability of culvert health assessment is improved, and structural failure caused by wooden barrel short plates is avoided.

In a specific embodiment, it is assumed that the detection is performed on a side wall of a long-distance underground culvert, and specific environment and detection parameters of the measuring point are as follows:

The environmental parameters are water depth h=10m, water temperature t=15deg.C, water flow velocity v=1.5m/s (assuming long-term service), and water density ρ _w=1000kg/m³. The resiliometer parameters are impact energy E ₀ = 2.207J, calibration parameters a=10, b=100. The underwater rebound measurement raw value R _w =17.5 measured at this point. At 0.3 meters near the point of measurement, there is a vertical slit with a width w=2 mm. The thickness of the member was 0.5m. The pressure correction coefficient k _p is calculated as the hydrostatic pressure p _w=ρ_w ×g×h=1000×9.8×10=98000 Pa. Assuming that the pressure influence coefficient λ _p=2×10^-6/Pa, k _p=1/(1+2×10^-6 ×98000) =1/1.196≡0.836. The temperature correction coefficient k _T is calculated assuming the concrete age is much greater than 90 days, k _age =1. Assuming that the temperature sensitivity coefficient α _T = -0.0004/° C, k _T = [1+ (-0.0004) × (15-20) ]×1=1+0.002=1.002. For simplicity of description, the provisional water flow scouring correction coefficient k _e and the medium coupling correction coefficient k _c according to the calibration result are k _e =0.98 and k _c =1.25, respectively. (in practical applications, the two terms should also be calculated by corresponding models), and the preliminary comprehensive coefficient k is calculated, wherein k=k _p×k_T×k_e×k_c =0.836×1.002×0.98×1.25≡1.025. The second order interaction correction term ak is calculated and k_final：Δk=0.05×(k_p-1)×(k_T-1)=0.05×(0.836-1)×(1.002-1)≈-0.0000164.k_final=k×(1+Δk)=1.025×(1-0.0000164)≈1.02498. is obtained, where k _final = 1.025 for clarity. preliminary dry equivalent strength f' _c：f'_c=k_final×(A×R_w -B) =1.025× (10×17.5-100) =1.025×75= 76.875MPa was calculated. The base intensity reduction coefficient k _crack is calculated assuming the crack critical width w_cr=5mm.k_crack=1-0.15×(w/w_cr)^0.5=1-0.15×(2/5)^0.5≈1-0.15×0.632=1-0.0948=0.9052. and the spatially varying defect correction coefficient k _d assuming the defect influencing radius σ is σ=0.5m. Distance r=0.3m of the measurement point to the defect center. Influence intensity I(r)=exp(-r²/2σ²)=exp(-0.3²/(2×0.5²))=exp(-0.09/0.5)=exp(-0.18)≈0.835. defect correction coefficient k _d=1-(1-k_crack) ×i (r) =1- (1-0.9052) ×0.835=1-0.0948 ×0.835≡1-0.0792 = 0.9208. The final corrected structural strength f _c：f_c=k_d×f'_c = 0.9208 × 76.875 ≡ 70.78MPa is calculated.

As can be seen from this example, the original underwater rebound reading 17.5, when not corrected, may correspond to a higher intensity value. But by multi-factor environmental correction (intensity is corrected from about 75MPa to 76.875 MPa) and key defect influence correction (intensity is further corrected from 76.875MPa to 70.78 MPa), the intensity value which is more in line with the actual health condition of the structure is obtained.

In another embodiment of the present application, a method for health assessment of a long-distance underground culvert structure includes:

Firstly, acquiring acoustic original data and an optical image sequence carried by an ROV, performing space-time fusion through propagation time difference compensation and partition self-adaptive correction, identifying and positioning structural defects, and outputting fusion detection data, defect characteristic maps and defect position coordinate sets with time-space alignment.

Specifically, acoustic waveform data (sampling rate is more than or equal to 100 kHz) and an optical image sequence (frame rate is more than or equal to 30 fps) of an ROV-mounted sensor are read, data integrity check and time stamp synchronization are carried out, and a time-stamped acoustic data stream and a time-stamped image data stream are obtained. And (3) reading the acoustic data stream with the time stamp and the image data stream with the time stamp, calculating the propagation speed difference (the sound speed is 1500m/s, the light speed is 2.25 multiplied by 10 ⁸ m/s) of sound and light in water, and performing time axis translation according to the sensor interval d to obtain time-aligned acoustic data and time-aligned optical data. And (3) reading a CAD drawing of the culvert structure, dividing the culvert into a top area, a side wall area and a bottom area, and calculating an acoustic attenuation coefficient alpha _i and an optical scattering coefficient beta _i of each area according to the material characteristics (concrete thickness and reinforcement distribution) and water flow conditions of each area to obtain a partition fusion parameter table. Reading the time aligned acoustic data, performing short-time Fourier transform to extract frequency domain features (0.1-50 kHz), identifying defect feature frequencies, and reading the time aligned optical data, performing edge detection and texture analysis to obtain an acoustic feature map and a visual feature map. And reading an acoustic feature map, a visual feature map and a partition fusion parameter table, applying weighted fusion F _i=α_i×A_i+β_i×V_i（A_i to each partition to obtain acoustic features and V _i to obtain a cross-modal feature verification to obtain a fusion feature map. Reading the fusion characteristic spectrum, carrying out threshold segmentation and connected domain analysis, identifying defects such as cracks (width >0.2 mm), seepage (wetting area >100cm ²), flaking (depth >5 mm) and the like, and calculating three-dimensional coordinates of the defects by combining ROV position data to obtain a defect position coordinate set, defect type labels and defect geometric parameters. Reading a fusion characteristic spectrum and an original acoustic characteristic spectrum, Calculating a fusion gain index G=SNR _fusion/(SNR_acoustic+SNR_visual according to the visual characteristic spectrum, wherein SNR _fusion is the signal-to-noise ratio of the sound-light characteristic after fusion, SNR _acoustic is the signal-to-noise ratio of the original acoustic signal, SNR _visual is the signal-to-noise ratio of the original optical image, and verifying the fusion effect (G >1.5 is effective fusion) to generate fusion detection data and a defect characteristic spectrum with time-space alignment.

And step two, reading a rebound measurement original value, coring test data and a defect position coordinate set, performing strength calculation through an underwater-dry land conversion model, and outputting a corrected concrete strength distribution diagram and a corrected material degradation parameter.

Specifically, the original rebound measurement value R _w (16 readings of each measuring point) of the underwater rebound instrument is read, the water depth h, the water temperature T and the water flow speed v are synchronously recorded, abnormal values (exceeding the mean value +/-3 sigma) are removed, and an effective rebound value set and an environment parameter set are obtained. And (3) reading a concrete core sample collected by the core drilling machine, performing laboratory compressive strength test, recording a core sample compressive strength value f _c, and comparing the effective rebound value sets at the same positions to establish a rebound value-strength corresponding relation so as to obtain a calibration data pair. And reading the calibration data pair and the environment parameter set, calculating a pressure correction term k _p =1-0.002 h according to the water pressure influence, calculating a temperature correction term k _T =1+0.01 (T-20) according to the temperature influence, and obtaining a conversion coefficient k=k _p×k_T×k₀（k₀ as a reference coefficient 1.15 in a comprehensive mode to obtain the underwater conversion coefficient. And (3) reading a defect position coordinate set and defect geometric parameters, applying strength reduction to rebound measuring points within a range of 50cm around the defect, namely 5-15% of crack reduction and 10-25% of peeling reduction according to the defect type, and calculating a correction coefficient k _d to obtain defect correction coefficient distribution. And (3) reading an effective rebound value set, an underwater conversion coefficient and a defect correction coefficient distribution, calculating the concrete strength f _c=k×k_d×(10×R_w -100 of each measuring point, and reconstructing the strength distribution among the measuring points to obtain a corrected concrete strength distribution map. And (3) reading the corrected concrete strength distribution diagram and the design strength value, calculating the strength loss rate eta= (f _design-f_current)/f_design, analyzing the strength gradient V.f to identify a degradation concentration area, and extracting the degradation rate v _d = delta f/delta t (based on historical data) to obtain a material degradation parameter, wherein f _design is the design strength value, f _current is the currently measured concrete strength value, and delta f is the strength variation.

And step three, reading defect characteristic maps, material degradation parameters and historical monitoring data, and outputting evolution state parameters, mutation risk indexes and evolution rate distribution through a continuous evolution model and a mutation detection algorithm.

The method comprises the steps of reading a defect characteristic map, identifying seepage traces and water flow paths, calculating a seepage area A _w and a wetting perimeter P _w, reading material degradation parameters, extracting porosity growth rate and permeability coefficient change, and obtaining a permeability characteristic parameter and an initial evolution state. And (3) reading historical monitoring data (including water level, flow, strain and the like), carrying out data cleaning and interpolation, unifying time resolution to 1 hour, and calculating hydraulic slope drop i, seepage velocity v _s and soil strain rate epsilon to obtain multi-physical-field time sequence data. And (3) reading multi-physical-field time sequence data and permeability characteristic parameters, calculating instantaneous evolution increment dphi/dt=alpha×Δi+beta×Δq+gamma×Δepsilon (weight coefficients alpha=0.4, beta=0.35 and gamma=0.25 are determined according to soil types), and performing time integration phi (t) = [ pi ] dphi/dt×dt to obtain an evolution degree time sequence. Reading an evolution degree time sequence, judging the current evolution stage according to a threshold value, wherein phi <0.2 is a water seepage stage, phi <0.2 is less than or equal to 0.4 is a permeation deformation stage, phi <0.6 is a permeation damage stage, phi <0.6 is less than or equal to 0.8 is a channel expansion stage, phi is more than or equal to 0.8 is a riverbed collapse stage, and obtaining the current evolution stage mark and stage conversion moment. Reading an evolution degree time sequence, calculating a first derivative psiphi/psit and a second derivative psi ²Φ/Ψt², calculating mutation sensitivity S _m=|Ψ²Φ/Ψt² I/psiphi/psit I, and marking as potential mutation points when S _m is more than 2.5 to obtain a mutation sensitivity curve and a potential mutation moment set. And (3) reading a time sequence and a potential mutation time set of the acoustic feature spectrum, analyzing the frequency spectrum changes before and after the mutation time, identifying high-frequency noise sudden increase (> 20kHz, and amplitude increase >10 dB) and low-frequency vibration (< 100 Hz) features, and confirming mutation types to obtain mutation risk indexes and mutation type labels. Optionally, obtaining mutation risk index and mutation type label includes reading acoustic feature map and potential mutation time set, extracting acoustic data in [ t _m-300s,t_m +300s ] time window for each mutation time t _m, performing short-time Fourier transform (window length 1024, overlapping 50%), calculating power spectral density PSD (f, t), dividing frequency band into low frequency (< 100 Hz), dividing frequency band, And calculating energy E _low、E_mid、E_high of each frequency band at medium frequency (100 Hz-10 kHz) and high frequency (more than 10 kHz) to obtain an acoustic feature matrix at the abrupt moment. The acoustic feature matrix at the abrupt change moment is read, statistics of feature time sequences are calculated, namely a high-frequency energy burst rate r _high=(E_{high_max}-E_{high_mean})/E_{high_mean}, wherein E _{high_max} is the maximum value of high-frequency band acoustic energy, E _{high_mean} is the average acoustic energy of the high-frequency band, low-frequency vibration intensity I _low=∫E_low (f <50 Hz) df is calculated, and abnormal modes are identified, namely water stopping tearing (r _high >3 and lasting for >10 s), a water stopping state and a water stopping state, Structural vibration (I _low is 5 times the mean value), particle scouring (2 kHz < f <5kHz energy increase >10 dB), and obtaining an acoustic abnormal mode label and an abnormal strength index. reading potential mutation time sets and acoustic abnormal mode labels, and calculating the time difference delta t _lag=t_acoustic-t_evolution between acoustic abnormal and evolution mutation, wherein t _acoustic is the central time point when a certain acoustic abnormal mode (such as high-frequency noise surge or low-frequency vibration) occurs, t _evolution is the time of a certain evolution mutation point identified in mutation sensitivity analysis, and typical time lags of different abnormal modes are counted, namely water stopping tearing (-60 s to +30 s), water stopping tearing, water stopping and water stopping, structural vibration (-120 s+180s), establishing an association strength index R _ae=exp(-|Δt_lag |/tau, wherein tau=60 s is the characteristic time, and obtaining an acoustic-mutation association matrix. reading mutation type and initially judging, The method comprises the steps of carrying out multi-source information fusion on an acoustic abnormal mode label and an acoustic-mutation association matrix, confirming mutation types when R _ae is more than 0.7 and abnormal modes are matched, calculating a comprehensive risk index R _risk=w₁×I_m+w₂×r_high+w₃×R_ae, wherein the weight w ₁=0.3、w₂=0.4、w₃=0.3,I_m is an acoustic abnormal strength index of a mutation event, carrying out risk classification, wherein R _risk is high risk, 0.5-0.8 is middle risk and <0.5 is low risk, and outputting mutation risk indexes and mutation type labels. And reading an evolution degree time sequence, a defect position coordinate set and a current evolution stage mark, calculating the local evolution rate v _Φ = dphi/dt of each defect point, and performing spatial interpolation according to the defect spacing and the hydraulic connectivity to obtain evolution rate distribution and evolution state parameters.

And step four, reading evolution state parameters, mutation risk indexes, evolution speed distribution and corrected concrete strength distribution map, and outputting a full-field health state distribution map, a structural health evaluation report and a grading early warning scheme through sparse reconstruction and comprehensive evaluation models.

Specifically, the evolution rate distribution and the defect position coordinate set are read, the evolution rate extreme points and the gradient change severe areas are identified, the positions of key monitoring points are determined, the distance between any two points is ensured to be less than 50m, all high-risk areas are covered, and the optimized monitoring point set is obtained. And (3) reading evolution state parameters of the optimized monitoring point set, and establishing an evolution propagation equation ²Φ-(1/c²)Ψ²Φ/Ψt² =S (x, y, z), wherein c is the evolution propagation speed (0.5-2 m/day), S is the source item intensity, and carrying out finite element solution to obtain the full-field evolution degree distribution phi (x, y, z, t). The full-field evolution degree distribution is obtained by reading a defect position coordinate set and evolution rate distribution, conducting grid encryption in a high gradient area (|phi| > 0.1/m), enabling the minimum unit size to be 0.5m, using sparse grids in an evolution slow area (|phi| < 0.01/m), enabling the maximum unit size to be 5m, generating unstructured grids through Delaunay triangulation, enabling grid quality indexes to be more than 0.7, and obtaining self-adaptive finite element grids and grid node coordinates. The method comprises the steps of reading an actual measurement evolution degree value phi _measured of an optimized monitoring point set, carrying out data quality inspection, removing abnormal points deviating from a mean value 3 sigma, adopting anisotropic kriging interpolation, adopting a variable range parameter of 50m in the longitudinal direction (along the water flow direction), 20m in the transverse direction and 10m in the vertical direction, calculating interpolation uncertainty sigma _kriging, marking a sparse area (the nearest monitoring point >30 m) of the monitoring point as low confidence, and obtaining an initial evolution degree field phi ₀ (x, y, z) and a confidence distribution map. Reading an initial evolution degree field, an evolution propagation equation coefficient field and an adaptive finite element grid to construct a discretization equation set, wherein [ M ] d ²Φ/dt² + [ C ] dphi/dt+ [ K ] phi= [ F ], wherein [ M ], [ C ], [ K ] are mass, damping and stiffness matrixes, [ F ] is an external load (source term) vector, an implicit time integration (Newmark-beta method, beta=0.25 and gamma=0.5) is adopted, a convergence criterion II phi ⁽ⁿ⁺¹⁾-Φⁿ‖/‖Φⁿ‖<10^-4 is set, wherein phi ⁽ⁿ⁺¹⁾ is an evolution degree value vector after n+1th step of iteration, and a space-time evolution degree field phi (x, y, z and t) is obtained through iterative solution. The method comprises the steps of reading actual measurement values of a space-time evolution degree field and an optimized monitoring point set, calculating a reconstruction error e _i=|Φ_{reconstructed}-Φ_measured|/Φ_measured, wherein phi _{reconstructed} is a reconstructed evolution degree value, carrying out cross verification, sequentially removing one monitoring point for reconstruction, calculating RMSE (root mean square of reconstruction error), analyzing uncertainty propagation, superposing a disturbance zone of +/-20% in a low confidence level area, and outputting full-field evolution degree distribution phi (x, y, z, t) and a reconstruction uncertainty cloud picture. And (3) reading the full-field evolution degree distribution, the corrected concrete strength distribution diagram and the mutation risk index, and calculating the comprehensive health index H=w ₁×(1-Φ)+w₂×(f_c/f_design)+w₃×(1-P_risk) and the weight w ₁=0.4、w₂=0.35、w₃ =0.25 to obtain the comprehensive health index distribution. The comprehensive health index distribution is read and divided according to five-level standard, wherein H is more than or equal to 0.85 and is more than or equal to 0.70 and is less than or equal to 0.85 and is basically healthy, H is more than or equal to 0.55 and is less than or equal to 0.70 and is sub-healthy, H is more than or equal to 0.40 and is less than or equal to 0.55 and is a disease, H is more than or equal to 0.40 and is a serious disease, and a color-coded full-field health state distribution diagram is generated. Reading evolution state parameters, evolution speed distribution and mutation risk indexes, establishing a time sequence prediction model, extrapolating evolution degree phi (t+deltat) of 30/90/180 days in the future, and identifying a time window which possibly exceeds a critical threshold value to obtain a risk evolution prediction curve and an early warning time node. Reading a full-field health state distribution diagram, early warning time nodes and mutation type labels, generating differential early warning according to risk grades, performing routine monitoring of blue early warning (within 180 days), performing encryption monitoring of yellow early warning (within 90 days), preparing maintenance of orange early warning (within 30 days), performing emergency treatment of red early warning (within 7 days), and outputting a grading early warning scheme. And integrating the full-field health status distribution map, the risk evolution prediction curve, the hierarchical early warning scheme and all key parameters to generate a structural health assessment report comprising current status diagnosis, evolution trend analysis, risk grade assessment and treatment suggestion.

The preferred embodiments of the present invention have been described in detail above, but the present invention is not limited to the specific details of the above embodiments, and various equivalent changes can be made to the technical solution of the present invention within the scope of the technical concept of the present invention, and all the equivalent changes belong to the protection scope of the present invention.

Claims (10)

1. The method for evaluating the health of the long-distance underground culvert structure is characterized by comprising the following steps of:

Based on the acoustic and optical detection data, fusion processing is carried out, and structural defect data comprising soil body types, defect types and geometric parameters are identified and output;

Based on pre-stored underwater rebound and coring test data, combining structural defect data, performing underwater-dry conversion and correction of strength values, and determining corrected structural strength and material degradation parameters;

combining the structural defect data, the material degradation parameters and the prestored historical monitoring data, carrying out structural evolution modeling and mutation risk identification, and obtaining the evolution characteristics and mutation risks of the structure;

and (3) comprehensively correcting the structural strength, evolution characteristics and mutation risk, and performing full-field state evaluation to generate structural health evaluation and risk early warning results.

2. The method of claim 1, wherein obtaining evolution features and mutation risk of the structure comprises:

Integrating the structural defect data, the material degradation parameters and the historical monitoring data, constructing and solving an evolution function of multi-physical field coupling, and generating an evolution degree time sequence of the accumulated evolution degree of the quantized structure;

Differential analysis is carried out on the evolution degree time sequence, and potential mutation moments representing risk transitions are identified;

deducing and generating an evolution rate distribution diagram according to the evolution degree time sequence and the spatial position of the structural defect data;

and forming the evolution characteristics and mutation risks of the structure together by the evolution degree time sequence, the potential mutation moment and the evolution rate distribution diagram.

3. The method of claim 2, wherein generating the evolution-degree time series comprises:

Extracting physical field indexes at least comprising hydraulic gradient, seepage and soil strain from historical monitoring data, and calculating respective time increment sequences;

dynamically calibrating evolution weight coefficients corresponding to the physical field indexes by referring to the soil body types and the material degradation parameters;

Carrying out weighted fusion on the time increment sequence by adopting an evolution weight coefficient to obtain an evolution increment sequence reflecting the instantaneous evolution rate of the structure;

And carrying out numerical integration on the evolution increment sequence along the time dimension to generate an evolution degree time sequence.

4. A method according to claim 3, wherein dynamically calibrating the evolution weight coefficients comprises:

According to the soil type, configuring basic weight coefficients for each physical field index;

extracting a structural degradation degree index from the material degradation parameters, and correcting at least one coefficient value in the basic weight coefficients according to the structural degradation degree index to obtain an adjusted weight;

And performing normalization calculation on the adjusted weights to generate evolution weight coefficients.

5. The method of claim 2, wherein identifying potential mutation moments that characterize a risk transition comprises:

solving a first derivative sequence and a second derivative sequence of the evolution degree time sequence;

constructing a mutation sensitivity time sequence by calculating the ratio of the absolute value of the corresponding value of the second derivative sequence to the absolute value of the corresponding value of the first derivative sequence at each moment;

And comparing each sensitivity value in the mutation sensitivity time sequence with a preset threshold value, and identifying the moment exceeding the threshold value as the potential mutation moment.

6. The method as recited in claim 5, further comprising:

extracting acoustic data segments in a preset time window before and after each potential mutation moment from the acoustic detection data aiming at each identified potential mutation moment;

Performing spectrum analysis on the acoustic data segment, and decoding whether an acoustic abnormal mode associated with a preset specific failure mechanism exists or not;

If the acoustic abnormal mode exists, the potential mutation moment is updated to be verified mutation risks, and the mutation risks are classified according to the acoustic abnormal mode.

7. The method of claim 1, wherein determining corrected structural strength and material degradation parameters comprises:

constructing and calibrating an underwater-dry land comprehensive conversion coefficient based on coring test data;

Converting the underwater rebound into a preliminary dry land equivalent intensity value by using an underwater-dry land comprehensive conversion coefficient;

and carrying out local correction on the preliminary dry equivalent strength value according to the defect type and the geometric parameter to obtain corrected structural strength, and generating a material degradation parameter based on the comparison between the corrected structural strength and the preset design strength.

8. The method of claim 7, wherein constructing and calibrating the underwater-dry land integrated transform coefficients comprises:

For physical effects of the underwater environment, independent environment correction factors are respectively quantized and determined, and the method at least comprises the following two steps:

A pressure correction coefficient characterizing the impact of hydrostatic pressure on rebound values;

a temperature correction coefficient for representing the influence of water temperature on the elastic modulus of the concrete;

The scouring correction coefficient which characterizes the influence of water flow scouring on the hardness of the concrete surface;

The medium coupling coefficient of the influence of the water medium on rebound impact energy transfer is represented;

and generating the underwater-dry land comprehensive conversion coefficient by coupling and integrating all the environmental correction factors.

9. The method of claim 7, wherein obtaining corrected structural strength comprises:

Aiming at each defect in the structural defect data, constructing a three-dimensional influence field model of which influence intensity decays along with the increase of the space distance according to the geometric parameters of the defects;

setting a basic strength reduction coefficient representing the maximum influence degree of the defect type based on the defect type;

fusing the three-dimensional influence field model and the basic strength reduction coefficient, and calculating a defect correction coefficient field which continuously changes in space;

And applying the defect correction coefficient field to the preliminary dry equivalent intensity value to obtain the corrected structural intensity.

10. The method according to claim 1, characterized in that the full-field state evaluation is performed, comprising a full-field reconstruction phase for generating a continuous state distribution, in particular:

Arranging evolution parameters contained in the evolution features and the mutation risks as sparse state monitoring points in the three-dimensional space of the culvert;

Constructing a physical constraint equation for describing the space-time evolution of the propagation rule of the evolution state in the structural medium;

and taking the data of the sparse state monitoring points as a solving boundary or an initial condition, carrying out numerical solving on a physical constraint equation, and generating a full-field evolution degree distribution map which continuously covers the whole culvert.