CN114997060A - A phononic crystal time-varying reliability testing method, computing device and storage medium - Google Patents

Abstract

The invention relates to the field of phononic crystals, and discloses a method for testing time-varying reliability of a phononic crystal, computing equipment and a storage medium, which comprises the following steps: determining a sample space according to parameters of the phononic crystal; constructing a reliability test model according to the sample space and the failure conditions of the phononic crystal; constructing a neural network model according to the reliability test model; and predicting the failure rate of the phononic crystal in the service time according to the neural network model and the sample space. According to the method, the failure rate of the phononic crystal in the service time can be predicted by constructing the reliability test model of the phononic crystal and constructing the neural network model through the reliability test model, the phenomenon that the performance of the phononic crystal cannot be judged due to the lack of parameters of the phononic crystal is avoided, and the performance of the phononic crystal in the service time is determined.

Description

A phononic crystal time-varying reliability testing method, computing device and storage medium

technical field

The invention relates to the field of phononic crystals, in particular to a time-varying reliability testing method, computing equipment and storage medium of phononic crystals.

Background technique

Phononic crystals generally refer to materials with different densities and elastic parameters that are compounded together in a periodic structure and have functional materials with elastic (acoustic) wave band gaps. However, the structure and material properties of phononic crystals lead to time-varying uncertainties in phononic crystals. Time-varying uncertainty means that one or more parameters of the phononic crystal change over time, resulting in fluctuations in the performance of the phononic crystal. Therefore, time-varying reliability testing of phononic crystals becomes particularly important.

When testing a phononic crystal in the prior art, the characteristic parameters of the phononic crystal are often set through sufficient sample data, and then a test model of the phononic crystal is constructed according to the set characteristic parameters to test the performance of the phononic crystal. However, in practical work and research, the sample data of phononic crystals are often difficult to obtain, and accurate characteristic parameters cannot be constructed, resulting in inaccurate test results.

To this end, a new time-varying reliability testing method for phononic crystals is required.

SUMMARY OF THE INVENTION

To this end, the present invention provides a time-varying reliability testing method for phononic crystals, so as to try to solve or at least alleviate the above problems.

According to one aspect of the present invention, there is provided a time-varying reliability testing method for phononic crystals, suitable for execution in a computing device, the method comprising the steps of: determining a sample space according to parameters of the phononic crystal; The reliability test model is constructed according to the failure conditions; the neural network model is constructed according to the reliability test model; the failure rate of the phononic crystal during the service time is predicted according to the neural network model and the sample space.

Optionally, in the method according to the present invention, the parameters of the phononic crystal include: random variable parameters, random process parameters, interval variable parameters and interval process parameters; a reliability test model is constructed according to the sample space and the failure conditions of the phononic crystal. The method includes the following steps: converting the random process parameters in the crystal parameters to obtain equivalent random variable parameters; converting the interval process parameters in the crystal parameters to obtain equivalent interval variable parameters; converting the time parameters of the service time of the phononic crystal into equivalent parameters Distribution time parameters; build reliability test models based on random variable parameters, equivalent random variable parameters, interval variable parameters, equivalent interval variable parameters, equivalent distribution time parameters and failure conditions.

Optionally, in the method according to the present invention, the failure condition includes: during the service time, when the lower limit of the band gap of the phononic crystal is greater than a preset frequency, the phononic crystal fails.

Optionally, in the method according to the present invention, constructing the neural network model according to the reliability test model includes the steps of: determining a training set according to the sample point set; training the neural network model according to the reliability test model and the training set.

Optionally, in the method according to the present invention, determining the training set according to the sample point set includes the steps of: transforming the sample points in the sample space to obtain the transformed sample point set; and determining the training set according to the transformed sample point set.

Optionally, in the method according to the present invention, determining the training set according to the transformed sample point set includes the steps of: determining a sample weight for each sample point in the transformed sample point set; Eigenvalues; determine the training set from the transformed sample point set according to the eigenvalues.

Optionally, in the method according to the present invention, transforming the sample points in the sample space includes the step of: performing equivalent uncertain transformation on each sample point in the sample space to obtain a time-independent sample point set.

Optionally, in the method according to the present invention, the method further includes the steps of: judging whether the failure rate of the instantaneous reliability calculated by the neural network model satisfies the stopping rule; A new sample set is determined according to the incremental sample points and the sample set; a new neural network model is trained according to the new sample set, until the neural network model that satisfies the stopping rule is obtained after training.

Optionally, in the method according to the present invention, determining the incremental sample points according to the number of iterations includes the steps of: determining a candidate point set according to the training set, and determining a first point set according to the candidate point set; sampling from the first point set through weight sampling Determine the second point set; determine incremental sample points from the second point set according to the active learning function.

Optionally, in the method according to the present invention, determining the incremental sample points from the second point set according to the active learning function includes the steps of: determining the uncertainty of each sample point in the second point set; determining each sample point in the second point set; The Euclidean distance between the sample point and the training set; input the uncertainty and Euclidean distance of each sample point into the active learning function to obtain the function value of each sample point; take the sample point with the smallest function value in the second point set as the incremental sample point.

Optionally, in the method according to the present invention, determining the uncertainty of each sample point in the second point set includes the steps of: generating a plurality of training complements according to the last generated sample set; generating a complement according to each training complement. Neural network model; determine the uncertainty of each sample point based on the supplementary neural network model and the last generated neural network model.

Optionally, in the method according to the present invention, predicting the failure rate of the phononic crystal within the service time according to the neural network model and the sample space includes the steps of: calculating and obtaining the phononic crystal according to the neural network model and the sample points in the sample space. Failure rates for hybrid time-varying reliability.

Optionally, in the method according to the present invention, the method further includes the steps of: calculating the coefficient of variation of the failure rate according to the failure rate of the mixed time-varying reliability of the phononic crystal; determining whether the coefficient of variation is greater than a preset coefficient threshold; if the coefficient of variation is not If it is greater than the preset coefficient threshold, new sample points are added to the sample space to obtain a new sample space; the neural network model is trained according to the new sample space until the coefficient of variation of the neural network model is greater than the preset minimization threshold.

According to another aspect of the present invention, there is provided a computing device comprising: one or more processors; a memory; and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by one or more A plurality of processors execute, and one or more programs include a method for executing the time-varying reliability testing method of a phononic crystal according to the present invention.

According to yet another aspect of the present invention, there is provided a computer-readable storage medium storing one or more programs, the one or more programs comprising instructions that, when executed by a computing device, cause the computing device to perform a program according to the present invention Time-varying reliability test method for phononic crystals.

The invention discloses a time-varying reliability testing method of a phononic crystal, which is suitable for execution in a computing device. The method includes the steps of: determining a sample space according to the parameters of the phononic crystal; building a reliability test model according to the sample space and the failure conditions of the phononic crystal; building a neural network model according to the reliability test model; predicting the phonon according to the neural network model and the sample space The failure rate of a crystal over its service time. By constructing the reliability test model of the phononic crystal and constructing the neural network model through the reliability test model, the invention can predict the failure rate of the phononic crystal within the service time, and avoid the inability to judge due to the lack of parameters of the phononic crystal. The performance of the phononic crystal realizes the determination of the performance of the phononic crystal during the service time.

Description of drawings

To achieve the above and related objects, certain illustrative aspects are described herein in conjunction with the following description and drawings, which are indicative of the various ways in which the principles disclosed herein may be practiced, and all aspects and their equivalents are intended to be within the scope of the claimed subject matter. The above and other objects, features and advantages of the present disclosure will become more apparent by reading the following detailed description in conjunction with the accompanying drawings. Throughout this disclosure, the same reference numbers generally refer to the same parts or elements.

FIG. 1 shows a schematic diagram of a time-varying reliability testing method 100 for phononic crystals according to an exemplary embodiment of the present invention;

FIG. 2 shows a structural block diagram of a computing device 200 according to an exemplary embodiment of the present invention;

Figure 3a shows a schematic diagram of acoustic wave propagation in a phononic crystal according to an exemplary embodiment of the present invention;

Figure 3b shows a schematic structural diagram of a phononic crystal according to an exemplary embodiment of the present invention;

4a-4d show schematic diagrams of equivalent uncertain transformation according to an exemplary embodiment of the present invention;

FIG. 5 shows a schematic diagram of a time-varying reliability analysis of a phononic crystal according to an exemplary embodiment of the present invention.

Detailed ways

Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be embodied in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided so that the present disclosure will be more thoroughly understood, and will fully convey the scope of the present disclosure to those skilled in the art. The same reference numbers generally refer to the same parts or elements.

The invention provides a time-varying reliability testing method for phononic crystals. The time-varying reliability testing method for phononic crystals is suitable for implementation in computing devices. FIG. 2 shows a structural block diagram of a computing device 200 according to an exemplary embodiment of the present invention.

In a basic configuration, computing device 200 includes at least one processing unit 220 and system memory 210 . According to one aspect, depending on the configuration and type of computing device, system memory 210 includes, but is not limited to, volatile storage (eg, random access memory), non-volatile storage (eg, read-only memory), flash memory, Or any combination of such memories. According to one aspect, system memory 210 includes an operating system 211 .

According to one aspect, operating system 211 , for example, is adapted to control the operation of computing device 200 . Furthermore, the examples are practiced in conjunction with graphics libraries, other operating systems, or any other application, and are not limited to any particular application or system. This basic configuration is shown in FIG. 2 by those components within dashed line 215 . According to one aspect, computing device 200 has additional features or functionality. For example, according to one aspect, computing device 200 includes additional data storage devices (removable and/or non-removable), such as magnetic disks, optical disks, or tapes.

As set forth above, according to one aspect, program modules 212 are stored in system memory 210 . According to one aspect, the program module 212 may include one or more applications. The present invention does not limit the types of applications, for example, applications also include: email and contacts applications, word processing applications, spreadsheet applications, database applications programs, slideshow applications, drawing or computer aided applications, web browser applications, etc.

According to one aspect, the examples may be practiced on a circuit comprising discrete electronic components, a packaged or integrated electronic chip containing logic gates, a circuit utilizing a microprocessor, or on a single chip containing electronic components or a microprocessor. For example, the examples may be practiced via a system-on-chip (SOC) in which each or many of the components shown in FIG. 2 may be integrated on a single integrated circuit. According to one aspect, such SOC devices may include one or more processing units, graphics units, communication units, system virtualization units, and various application functions, all integrated (or "burned") into a single integrated circuit on the chip substrate. When operating via the SOC, the functions described herein may operate via dedicated logic integrated with other components of the computing device 200 on a single integrated circuit (chip). Embodiments of the invention may also be practiced using other technologies capable of performing logical operations such as AND, OR, and NOT, including but not limited to mechanical, optical, fluid, and quantum technologies. Additionally, embodiments of the invention may be practiced within a general purpose computer or in any other circuit or system.

According to one aspect, computing device 200 may also have one or more input devices 231, such as a keyboard, mouse, pen, voice input device, touch input device, and the like. Output devices 232 may also be included, such as displays, speakers, printers, and the like. The aforementioned devices are examples and other devices may also be used. Computing device 200 may include one or more communication connections 233 that allow communication with other computing devices 200 . Examples of suitable communication connections 233 include, but are not limited to: RF transmitter, receiver and/or transceiver circuits; Universal Serial Bus (USB), parallel and/or serial ports.

Embodiments of the present invention also provide a non-transitory readable storage medium storing instructions for causing the computing device to perform a method according to an embodiment of the present invention. The readable medium of this embodiment includes permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information may be computer readable instructions, data structures, modules of programs, or other data. Examples of readable storage media include, but are not limited to: phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read only memory (ROM), Electrically Erasable Programmable Read Only Memory (EEPROM), Flash Memory or other memory technology, Compact Disc Read Only Memory (CD-ROM), Digital Versatile Disc (DVD) or other optical storage, Magnetic tape cartridges, magnetic tape disk storage or other magnetic storage devices or any other non-transitory readable storage medium.

According to one aspect, communication media are embodied by computer readable instructions, data structures, program modules, or other data in a modulated data signal (eg, carrier wave or other transport mechanism) and include any information delivery media. According to one aspect, the term "modulated data signal" describes a signal that has one or more sets of characteristics or is altered in such a way as to encode information in the signal. By way of example and not limitation, communication media includes wired media such as a wired network or direct wired connection, as well as wireless media such as acoustic, radio frequency (RF), infrared, and other wireless media.

It should be noted that, although the above computing device only shows the processing unit 220, the system memory 210, the input device 231, the output device 232, and the communication connection 233, in the specific implementation process, the device may also include the Additional components required. In addition, those skilled in the art can understand that, the above-mentioned device may only include components necessary to implement the solutions of the embodiments of the present specification, rather than all the components shown in the figures.

Next, the specific execution process of the time-varying reliability testing method of the phononic crystal of the present invention will be described in detail according to FIG. 1 . FIG. 1 shows a schematic diagram of a time-varying reliability testing method 100 of a phononic crystal according to an exemplary embodiment of the present invention. As shown in FIG. 1 , step S110 is first performed to determine the sample space according to the parameters of the phononic crystal.

In phononic crystals, materials with different densities and elastic parameters related to the propagation of elastic waves are periodically combined according to the structure similar to natural crystals. The materials distributed on the lattice points that are not connected to each other are called scatterers, and the connected ones are integrated as a whole. The background dielectric material is called the matrix. The so-called elastic wave forbidden band means that there is no elastic wave eigenmode in a certain frequency range, that is, elastic waves in this frequency range are prohibited from propagating.

Phononic crystals can be divided into three types: one-dimensional (layered), two-dimensional and three-dimensional phononic crystals according to the shape of the scatterer and the form of its periodic distribution in the matrix. According to the periodicity of the lattice, common two-dimensional phononic crystals are divided into: square lattice, triangular lattice, hexagonal lattice and so on. According to an embodiment of the present invention, the phononic crystal used in the time-varying reliability test of the present invention is specifically a two-dimensional phononic crystal.

Figure 3a shows a schematic diagram of acoustic waves propagating in a phononic crystal according to an exemplary embodiment of the present invention.

The structure and material properties of phononic crystals have various parameters, which can be divided into random variable parameters, random process parameters, interval variable parameters and interval process parameters according to the continuity and time correlation.

According to an embodiment of the present invention, the random variable parameter includes Young's modulus and density of the scatterer and the matrix; the random process parameter includes the side length of the matrix; the interval variable parameter includes the Poisson's ratio of the scatterer and the matrix; the interval process parameter includes The diameter of the scatterer.

FIG. 3b shows a schematic structural diagram of a phononic crystal according to an exemplary embodiment of the present invention. As shown in Fig. 3b, the phononic crystal includes a scatterer and a matrix; wherein, the diameter of the scatterer is _RC , and the length of the matrix is a.

According to the value range of each parameter of the phononic crystal, the sample space for the parameters of the phononic crystal to be valued can be determined. The sample space includes multiple sample points. Each sample point includes multiple parameters, and each parameter falls within the value range of the corresponding parameter of the phononic crystal. Each sample point in the sample space represents a possible parameter value of the phononic crystal.

According to an embodiment of the present invention, in order to simulate the parameter values of the phononic crystal, Monte Carlo simulation can be used to randomly distribute the sample points in the sample space. At this time, each sample point in the sample space is a Monte Carlo sample point. The number of sample points in the sample space is N _mc , and the set of sample points formed by the N _mc sample points is the sample point set S _MCS .

Subsequently, step S120 is performed to construct a reliability test model according to the sample space and the failure conditions of the phononic crystal.

According to an embodiment of the present invention, in order to explore the reliability of the phononic crystal, a reliability test model needs to be constructed to test the phononic crystal. The basic rule of the reliability test model is that when the phononic crystal is in service time and the lower limit of the band gap of the elastic wave forbidden band is less than the preset frequency, the phononic crystal structure is considered to be invalid. When the lower limit of the band gap of the elastic wave forbidden band of the phononic crystal is smaller than the preset frequency, the hindering performance of the phononic crystal to elastic waves is not as expected, and the phononic crystal fails. Therefore, the failure condition of the phononic crystal is: during the service time, when the lower limit of the band gap of the phononic crystal is greater than the preset frequency, the phononic crystal fails. According to the failure conditions of the phononic crystal, the reliability test model of the phononic crystal can be established.

其中，X＝(X₁，X₂，...X_m)，表示由随机变量参数组成的m维向量；Y＝(Y₁，Y₂，...Y_n)，表示由区间变量参数组成的n维向量；S(t)＝[S₁(t)，S₂(t)，...S_k(t)]，表示由k个随机过程参数组成的向量；I(t)＝[I₁(t)，I₂(t)，...I_l(t)]，表示由l个区间过程参数组成的向量；t为声子晶体预设的服役时间。According to an embodiment of the present invention, the lower limit of the band gap of the phononic crystal can be calculated by the finite element method. The lower limit of the band gap is calculated by random variable parameters, random process parameters, interval variable parameters and interval process parameters. The specific calculation results are:

Wherein, X=(X ₁ , X ₂ ,...X _m ), represents the m-dimensional vector composed of random variable parameters; Y=(Y ₁ , Y ₂ ,... Y _n ), represents the interval variable parameters composed of n-dimensional vectors; S(t)=[S ₁ (t), S ₂ (t),...S _k (t)], representing a vector composed of k random process parameters; I(t)= [I ₁ (t), I ₂ (t),...I _l (t)], represents a vector composed of l interval process parameters; t is the preset service time of the phononic crystal.

The preset frequency f _th can be set as required. The present invention does not limit the specific value of the preset frequency f _th . According to an embodiment of the present invention, the preset frequency f _th may be set to 268.

According to the failure conditions of the phononic crystal, the limit state equation for calculating whether the phononic crystal fails can be expressed as:

Among them, g(X, Y, S(t), I(t), t) represents the interpolation between the preset frequency and the lower limit of the band gap of the phononic crystal, when g(X, Y, S(t), I(t) ), when t) is less than 0, the phononic crystal fails.

The time-varying declaration period of the phononic crystal is [0, T _L ], and the service time of the phononic crystal within the time-varying declaration period is T, _0≤T≤TL . The failure probability of phononic crystals in service time can be expressed as:

When the parameters of the phononic crystal take specific values, the maximum and minimum values of the lower bound of the band gap are calculated. Correspondingly, the lower bound of the maximum band gap is substituted into the limit state equation, and the lower bound of the failure probability of the phononic crystal is calculated; the lower bound of the minimum band gap is substituted into the limit state equation to calculate the upper bound of the failure probability of the phononic crystal. The lower and upper bounds of the failure rate of phononic crystals in service time can be expressed as:

In order to facilitate the reliability testing of phononic crystals and reduce the influence of service time on random process parameters and interval process parameters, the equivalent uncertainty transformation is applied to the random process parameters and interval process parameters to obtain the time-independent equivalent Random variable parameters and equivalent interval variable parameters, thereby transforming time-varying reliability analysis into time-invariant reliability analysis.

According to an embodiment of the present invention, the equivalent uncertain transformation can be specifically realized by the following formula:

Among them, P _t =[S(t), I(t)], P _t represents time-varying uncertain parameters, including S(t) and I(t). P _t '=[S', I'], P _t ' represents the time-invariant uncertain parameters after transformation, including S' and I'. S' is the equivalent random variable parameter transformed from the random process parameter, and I' is the equivalent interval variable parameter transformed from the interval process parameter.

4a-4d illustrate schematic diagrams of equivalent indeterminate transformations according to an exemplary embodiment of the present invention. Figures 4a to 4d illustrate the process of equivalent uncertainty conversion by taking the conversion of interval process parameters into equivalent interval variable parameters as an example.

As shown in Figure 4a, the interval process parameter I(t) is related to time t; the upper bound function ^IU (t) and the lower bound function ^IL (t) determine the value range of the interval process parameter I(t) at time t. The time t is continuous time, and its value range is [0, T].

As shown in Fig. 4b, firstly, N discrete times will be taken in the continuous time t: t ₁ ˜t _N . Determine the value of the upper bound function I ^U (t) in discrete time, and obtain the discrete upper bound function I ^U (t _i ) of the interval process parameter I(t). Determine the value of the lower bound function ^IL (t) at discrete time ti, and obtain the discrete lower bound function ^IL (t _i ) of the interval process parameter I(t). Wherein, i is a positive integer between 1 and N, and i=1, 2, ..., N.

As shown in Figure 4c, the values of the discrete upper bound function and the discrete lower bound function at the discrete time t _i determine the value range of the interval process parameter I(t) at the discrete time t _i . Then, the value range of interval process parameter I(t) at discrete time t _i is taken as interval variable I _i , where i is a positive integer between 1 and N, i=1, 2,  , N.

As shown in Fig. 4d, the interval variables I ₁ to I _N are counted to determine the value probability function f _PDF (I) of the interval process parameter I(t) on the value space. The value probability function f _PDF (I) is the time-independent equivalent interval variable parameter.

According to an embodiment of the present invention, the time parameter t describing the service time of the phononic crystal is also transformed into an equivalent distribution time parameter t' equally distributed in the time-varying life cycle [0, T _L ] of the phononic crystal, t'˜U(0, T _L ).

The reliability test model can be constructed according to the parameters of random variables, interval variables, and the transformed equivalent random variable parameters, equivalent interval variable parameters and equivalent distribution time parameters.

In the final reliability test model, the limit state function used to calculate whether the phononic crystal fails is:

The corresponding failure probability of phononic crystals calculated by the reliability test model can be calculated by the following formula:

P _f =Pr(g(X, Y, S', I', t')<0)

Next, step S130 is executed to construct a neural network model according to the reliability test model.

When the failure probability of the phononic crystal needs to be predicted according to the reliability test model, the neural network model is trained according to the reliability test model to predict the failure probability of the phononic crystal. According to an embodiment of the present invention, the adopted neural network model may specifically be a DNN neural network model, and the present invention does not limit the specific type of the neural network model.

以便在训练时减少时间对随机过程参数和区间过程参数的影响。随后，从生成的样本点集

中选择样本点，得到训练集S。训练集S中包括的样本点的数目为M，M≤N_mc，N_mc为样本点集

中所包括样本点的数目。本发明对训练集S中所包括样本点的具体数目不做限制，可综合考虑训练时间等因素按需确定。When training the neural network model, the equivalent uncertainty transformation is firstly performed on the sample point set S _MCS that constitutes the sample space, and the time-related parameters of the sample points in the sample point set are converted into time-independent parameters according to the formula of equivalent uncertainty transformation. , get the transformed sample point set

In order to reduce the effect of time on random process parameters and interval process parameters during training. Subsequently, from the generated sample point set

Select the sample points in , and get the training set S. The number of sample points included in the training set S is M, M≤N _mc , and N _mc is the sample point set

The number of sample points included in . The present invention does not limit the specific number of sample points included in the training set S, and can be determined as needed by comprehensively considering factors such as training time.

According to an embodiment of the present invention, when sample points are selected from the sample space to obtain the training set S, the sample points may be selected by means of weighted sampling. Weighted sampling is used to solve the problem of sampling imbalance. Since random sampling collects more samples in the area with high sampling space probability, and the sample points near the limit state surface are generally in the area with small sampling space probability, in order to ensure that the samples obtained by sampling are as uniform as possible, weighted sampling is used to give small samples to small samples. The probability sample points are heavily weighted to ensure that the samples obtained by sampling are as uniform as possible.

When using weight sampling to obtain the training set S from the sample space, first calculate the sample weight of the sample point V _t ⁽ⁱ⁾ in the sample space, which can be calculated by the following formula:

Among them, w ⁽ⁱ⁾ is the sample weight of the sample point, and f(V _t ⁽ⁱ⁾ ) is the probability density of the sample point V _t ⁽ⁱ⁾ in the sample space. The parameter of the sample point V _t ⁽ⁱ⁾ is V _t =[X, Y, S', I', t'].

Then, the eigenvalue of each sample point is calculated according to the sample weight, which is calculated by the following formula:

Among them, u ⁽ⁱ⁾ can be set to a random number in the range of (0, 1). The present invention does not limit the specific value range of u ⁽ⁱ⁾ , which can be specifically set as required.

Finally, sort each sample point in ascending order according to the eigenvalue, and select the first M number of sample points to obtain the training set S.

中，未被选中的样本点，也即不在训练集S中的样本点作为候选点集S^*。候选点集S^*与训练集S组合能够得到样本点集

样本点集

中，候选点集S^*是训练集S的补集。sample point set

, the unselected sample points, that is, the sample points not in the training set S, are used as the candidate point set S ^* . The candidate point set S ^* can be combined with the training set S to obtain the sample point set

sample point set

, the candidate point set S ^* is the complement of the training set S.

输入训练好的神经网络模型，得到声子晶体的瞬时可靠性的失效率。After the training set S is obtained, the neural network model is trained according to the training set S. After training the neural network model, the failure rate of the phononic crystal is predicted according to the neural network model. According to an embodiment of the present invention, when predicting the failure rate of phononic crystals, the transformed sample point set is

Input the trained neural network model to obtain the instantaneous reliability failure rate of the phononic crystal.

According to an embodiment of the present invention, in order to test the reliability of phononic crystals more accurately, the present invention uses an iterative learning method to update the neural network model obtained by training, so as to obtain more accurate reliability of phononic crystals. Failure Rate.

According to one embodiment of the present invention, a stopping rule is set so as to determine when to end the iteration according to the stopping rule, and stop updating the neural network model. After each iteration is performed, after a new neural network model is determined, the failure rate of the phononic crystal is calculated according to the new neural network model, and then it is judged whether to stop the iteration according to the stopping rule. If it is judged that the stopping rule is not satisfied, continue to iterate and update the application network model. If it is judged that the stopping rule is satisfied, the iteration is stopped and the loop is exited.

Stopping rules include:

表示第i次迭代过程中，神经网络模型计算的声子晶体的失效率，n≤i≤k。ε_th为自定义停止阈值，可考虑测试时间等因素按需进行确定。根据本发明的一个实施例，ε_th可设置为0.01。n为自定义的需要计算的迭代范围；判断停止规则时，根据第n次到第k次迭代的声子晶体的失效率计算是否满足停止规则。Among them, k represents the k-th iteration process, n represents the n-th iteration process, and 1≤n≤k.

Indicates the failure rate of the phononic crystal calculated by the neural network model during the ith iteration, n≤i≤k. ε _th is a custom stop threshold, which can be determined as needed considering factors such as test time. According to an embodiment of the present invention, ε _th may be set to 0.01. n is a user-defined iteration range that needs to be calculated; when judging the stopping rule, calculate whether the stopping rule is satisfied according to the failure rate of the phononic crystal from the nth to the kth iteration.

输入训练好的神经网络模型，得到声子晶体的瞬时可靠性的失效率。接着判断是否满足停止规则；若满足停止规则，则停止迭代。随后，根据将样本点集S_MCS输入初始生成的神经网络模型，输出声子晶体的混合时变可靠性的失效率。若不满足停止规则，则继续迭代，对神经网络模型进行更新。According to an embodiment of the present invention, the initial generation of the neural network model is recorded as the first iteration, and the number of iterations k=1, and then the transformed sample point set is

Input the trained neural network model to obtain the instantaneous reliability failure rate of the phononic crystal. Then it is judged whether the stopping rule is satisfied; if the stopping rule is satisfied, the iteration is stopped. Then, according to the input of the sample point set S _MCS into the initially generated neural network model, the failure rate of the hybrid time-varying reliability of the phononic crystal is output. If the stopping rule is not satisfied, continue to iterate and update the neural network model.

具体的：使用上一次生成的神经网络模型，对候选点集S^*中每个样本点的进行计算得到每个样本点的响应值g，响应值g根据如下公式计算：When updating the neural network model, first determine the first point set from the candidate point set S ^*

Specifically: using the neural network model generated last time, calculate each sample point in the candidate point set S ^* to obtain the response value g of each sample point, and the response value g is calculated according to the following formula:

根据本发明的一个实施例，

的大小可通过如下公式计算：Then sort according to the absolute value of the response value of each sample point, and select N _S ^* sample points as the first point set

According to an embodiment of the present invention,

The size can be calculated by the following formula:

的具体取值不作限制，可具体根据训练需要进行确定。the present invention

The specific value of is not limited, and can be determined according to the training needs.

中按照权重采样的方式选择M个实验点，作为第二点集S_K。本发明对M的具体取值不做限制，可根据需要进行确定。Then, from the first point set

M experimental points are selected according to the weight sampling method as the second point set S _K . The present invention does not limit the specific value of M, which can be determined as required.

Next, an incremental sample point V _new is determined from the second point set _SK using an active learning function.

Specifically: First, the K-fold validation cross-validation method is used to process the training set. The training set is equally divided into k training subsets, and each training subset includes the same number of sample points. Then, during each training, one training subset is selected from the k training subsets as a test subset, and the other training subsets are used as training complements to train the neural network model. Taking each training subset in the k training subsets as a test subset in turn, k complementary neural network models can be obtained by training.

其中，l为1～k间的正整数，l＝1、2、......、k。Input the sample points into the supplementary neural network model to get the response value g. The supplementary neural network model is obtained by training according to the training supplement

Wherein, l is a positive integer between 1 and k, and l=1, 2, ..., k.

The response value is calculated according to the following formula:

Input each sample point in the second point set SK into the neural network model trained last time using the training set and _k complementary neural network models trained using the training complement set, and calculate the uncertainty of each sample point. sex. Specifically, it can be calculated by the following formula:

为将样本点输入到上一次使用训练集训练的神经网络模型计算得到的响应值，

为将样本点输入到第l个补充神经网络模型计算得到的响应值。Among them, us(V _t ⁱ ) is the uncertainty of each sample point in the second point set _SK .

The response value calculated for inputting the sample points to the neural network model last trained using the training set,

Response value computed for inputting sample points into the lth complementary neural network model.

Then, calculate the Euclidean distance between each sample point in the second point set SK and the existing training set _S , which can be calculated by the following formula:

Among them, d(V _t ⁱ ) is the Euclidean distance between each sample point and the existing training set _S , V _t ⁱ is the sample point in the second point set SK, i is a positive integer between 1 and M, and i = 1, 2, ..., M. V _t ^j is a sample point in the existing training set S, j is a positive integer between 1 and M, and j=1, 2, ..., M.

Finally, use the active learning function to determine the incremental sample point V _new according to the Euclidean distance and uncertainty, which can be calculated by the following formula:

Among them, _N (·) represents the normalization operation, and SLF(V _t ⁱ ) is the function value calculated by inputting each sample point in the second point set SK into the active learning function. The value range of β is (0, 1), and the present invention does not limit the specific value of β. According to an embodiment of the present invention, the value of β may be 0.5, which means that Euclidean distance and uncertainty are comprehensively considered with equal importance when calculating the active learning function.

Sort the function values obtained by inputting the active learning function to each sample point in the second point set _SK , and use the sample point with the smallest function value obtained as the incremental sample point V _new .

Subsequently, the incremental sample points are added to the sample set S to obtain a new sample set, and a new neural network model is obtained by training.

According to the above steps, the neural network model is iteratively updated by using the active learning method, and each time the most suitable incremental sample points are selected and added to the sample set to obtain a new sample set. The incremental sample points calculated according to the above method are far away from the existing sample points, which prevents the model from having problems, ensures the progress of the model, reduces the number of training sample points, and makes the sample points as close to the limit state surface as possible, and also has a high uncertainty. certainty.

The incremental sample points are added to the sample set S to obtain a new sample set, and at the same time, the incremental sample points are removed from the candidate point set S ^* to obtain a new candidate point set for subsequent iterations.

输入新的神经网络模型，得到声子晶体的瞬时可靠性的失效率。接着判断是否满足停止规则；若满足停止规则，则停止迭代。After training a new neural network model, the sample point set is

Input the new neural network model to get the failure rate of the transient reliability of the phononic crystal. Then it is judged whether the stopping rule is satisfied; if the stopping rule is satisfied, the iteration is stopped.

Finally, step S140 is performed to predict the failure rate of the phononic crystal within the service time according to the neural network model and the sample space.

According to the input of the sample point set S _MCS into the new neural network model, the failure rate of the hybrid time-varying reliability of the phononic crystal is output. If the stopping rule is not satisfied, continue to iterate and increase the number of iterations k by 1; then continue to update the newly generated neural network model until the failure rate of the instantaneous reliability of the output satisfies the stopping rule.

According to an embodiment of the present invention, after the failure rate of the mixed time-varying reliability of the phononic crystal is calculated, the coefficient of variation of the failure rate is calculated according to the failure rate, which can be specifically calculated by the following formula:

Among them, P _f is the failure rate, N is the number of sample points in the sample space, and specifically, it can be implemented as the Monte Carlo sampling number N _mc .

Then, the calculated coefficient of variation is compared with the preset coefficient threshold, and if the coefficient of variation is greater than the preset coefficient threshold, the failure rate is output. According to an embodiment of the present invention, the preset coefficient threshold may be set to 0.05, and the present invention does not limit the specific value of the preset coefficient threshold.

If the coefficient of variation is less than the preset coefficient threshold, increase the number of sample points in the sample space to obtain a new sample space, and execute the above steps S110 to S1n0 according to the new sample space to obtain a new neural network model, calculate the failure rate and coefficient of variation until the calculated coding coefficient is greater than the preset coefficient threshold.

和上界

因此在最终生成失效率包括失效率上界和失效率下界，失效率即在失效率上界和实效率下界之间。In the reliability test model, when calculating the failure rate of phononic crystals in service time, the lower bound is included

and upper bound

Therefore, the final generation failure rate includes the upper bound of the failure rate and the lower bound of the failure rate, and the failure rate is between the upper bound of the failure rate and the lower bound of the actual rate.

FIG. 5 shows a schematic diagram of a time-varying reliability analysis of a phononic crystal according to an exemplary embodiment of the present invention. As shown in Figure 5, the parameters of the phononic crystal are first determined according to its own structure and material properties; these parameters include: random variable parameters, random process parameters, interval variable parameters and interval process parameters. According to each parameter, the specific value range of each parameter can determine the sample space of the phononic crystal parameter value. The sample space includes multiple sample points. When taking sample points in the sample space, Monte Carlo can be used to generate N _mc sample points, and the set formed by these sample points is the sample point set S _MCS .

Then build a reliability test model to test the phononic crystal. When building the reliability test model, when setting the reliability test model, the failure condition of the phononic crystal is determined as follows: within the service time, when the lower limit of the band gap of the phononic crystal is greater than the preset frequency, the phononic crystal fails. The failure conditions of phononic crystals are used as the basic rules for building reliability test models, so as to build reliability test models.

In the reliability test model: the limit state equation for calculating whether the phononic crystal fails can be expressed as:

According to the limit state equation, the failure probability of the phononic crystal within the service time can be further calculated as:

When the parameters of the phononic crystal take specific values, the maximum and minimum values of the lower bound of the band gap are calculated. Correspondingly, the lower bound of the maximum band gap is substituted into the limit state equation, and the lower bound of the failure probability of the phononic crystal is calculated; the lower bound of the minimum band gap is substituted into the limit state equation to calculate the upper bound of the failure probability of the phononic crystal. The lower and upper bounds of the failure rate of phononic crystals in service time can be expressed as:

In order to reduce the influence of service time on random process parameters and interval process parameters when testing the reliability of phononic crystals, the equivalent uncertainty transformation is applied to the random process parameters and interval process parameters. The specific formula is as follows:

After the transformation, the time-independent equivalent random variable parameters and equivalent interval variable parameters are obtained, thereby transforming the time-varying reliability analysis into a time-invariant reliability analysis.

The time parameter t describing the service time of the _phononic crystal is also transformed into the equivalent distribution time parameter t', t'～U(0, T _L ).

The reliability test model can be constructed according to the parameters of random variables, interval variables, and the transformed equivalent random variable parameters, equivalent interval variable parameters and equivalent distribution time parameters.

In the reliability test model, the limit state function used to calculate whether the phononic crystal fails is correspondingly modified as:

The formula for calculating the failure probability of a phononic crystal correspondingly becomes:

P _f =Pr(g(X, Y, S', I', t')<0)

When the failure probability of the phononic crystal needs to be predicted according to the reliability test model, the neural network model is trained according to the reliability test model to predict the failure probability of the phononic crystal. In order to test the reliability of the phononic crystal more accurately, the present invention uses an iterative learning method to update the neural network model obtained by training. Therefore, when training the neural network model for the first time, set the number of iterations to 1.

随后，从生成的样本点集

中选择样本点，得到训练集S。训练集S中包括的样本点的数目为M，M≤N_mc，N_mc为样本点集

中所包括样本点的数目。When training the neural network model, the equivalent uncertainty transformation is firstly performed on the sample point set S _MCS that constitutes the sample space, and the time-related parameters of the sample points in the sample point set are converted into time-independent parameters according to the formula of equivalent uncertainty transformation. , get the transformed sample point set

Subsequently, from the generated sample point set

Select the sample points in , and get the training set S. The number of sample points included in the training set S is M, M≤N _mc , and N _mc is the sample point set

The number of sample points included in .

When selecting sample points from the sample space to obtain the training set S, the weight sampling method is used to select the sample points. First, calculate the sample weight of the sample point V _t ⁽ⁱ⁾ in the sample space, which can be calculated by the following formula:

Then, the eigenvalue of each sample point is calculated according to the sample weight, which is calculated by the following formula:

Finally, sort each sample point in ascending order according to the eigenvalue, and select the first M number of sample points to obtain the training set S.

中，未被选中的样本点，不在训练集S中的样本点作为候选点集S^*。sample point set

, the sample points that are not selected, and the sample points that are not in the training set S are taken as the candidate point set S ^* .

输入训练好的神经网络模型，得到声子晶体的瞬时可靠性的失效率。After the training set S is obtained, the neural network model is trained according to the training set S. After training the neural network model, the failure rate of the phononic crystal is predicted according to the neural network model;

Input the trained neural network model to obtain the instantaneous reliability failure rate of the phononic crystal.

Then, according to the stopping rule, it is judged whether the obtained failure rate of the instantaneous reliability of the phononic crystal satisfies the stopping rule. Stopping rules include:

If the stopping rule is satisfied, the sample point set S _MCS is input into the neural network model, and the failure rate of the mixed time-varying reliability of the phononic crystal is output.

If the stopping rule is not satisfied, update the number of iterations to increase the number of iterations by 1.

使用上一次生成的神经网络模型，对候选点集S^*中每个样本点的进行计算得到每个样本点的响应值g，响应值g根据如下公式计算：Subsequently, the first point set is determined from the candidate point set S ^*

Using the neural network model generated last time, calculate each sample point in the candidate point set S ^* to obtain the response value g of each sample point, and the response value g is calculated according to the following formula:

个样本点作为第一点集

的大小可通过如下公式计算：Then sort according to the absolute value of the response value of each sample point, select

sample points as the first point set

The size can be calculated by the following formula:

中按照权重采样的方式选择M个实验点，作为第二点集S_K。Then, from the first point set

M experimental points are selected according to the weight sampling method as the second point set S _K .

Next, an incremental sample point V _new is determined from the second point set _SK using an active learning function. First, the K-fold validation cross-validation method is used to process the training set. The training set is equally divided into k training subsets, and each training subset includes the same number of sample points. Each training subset in the k training subsets is used as a test subset in turn, and the other training subsets are used as training complements to train the neural network model, so that k complementary neural network models can be obtained by training.

响应值根据如下公式计算：Input the sample points into the supplementary neural network model to get the response value g. The supplementary neural network model is obtained by training according to the training supplement

The response value is calculated according to the following formula:

Input each sample point in the second point set SK into the neural network model trained last time using the training set and _k complementary neural network models trained using the training complement set, and calculate the uncertainty of each sample point. sex. Specifically, it can be calculated by the following formula:

Then, calculate the Euclidean distance between each sample point in the second point set SK and the existing training set _S , which can be calculated by the following formula:

Finally, use the active learning function to determine the incremental sample point V _new according to the Euclidean distance and uncertainty, which can be calculated by the following formula:

Sort the function values obtained by inputting the active learning function to each sample point in the second point set _SK , and use the sample point with the smallest function value obtained as the incremental sample point V _new .

Add incremental sample points to the sample set S to obtain a new sample set, and train to obtain a new neural network model.

输入新的神经网络模型，得到声子晶体的瞬时可靠性的失效率。接着判断是否满足停止规则；若满足停止规则，则停止迭代。随后，根据将样本点集S_MCS输入新的神经网络模型，输出声子晶体的混合时变可靠性的失效率。若不满足停止规则，则继续迭代，令迭代次数k增加1；接着对新生成的神经网络模型继续进行更新，直到输出的瞬时可靠性的失效率满足停止规则。After training a new neural network model, the sample point set is

Input the new neural network model to get the failure rate of the transient reliability of the phononic crystal. Then it is judged whether the stopping rule is satisfied; if the stopping rule is satisfied, the iteration is stopped. Then, according to the input of the sample point set S _MCS into a new neural network model, the failure rate of the hybrid time-varying reliability of the phononic crystal is output. If the stopping rule is not satisfied, continue to iterate and increase the number of iterations k by 1; then continue to update the newly generated neural network model until the failure rate of the instantaneous reliability of the output satisfies the stopping rule.

When the stopping rule is satisfied, the coefficient of variation of the failure rate is calculated according to the obtained failure rate of the mixed time-varying reliability of the phononic crystal, which can be calculated by the following formula:

The calculated coefficient of variation is compared with the preset coefficient threshold, and if the coefficient of variation is greater than the preset coefficient threshold, the failure rate is output.

If the coefficient of variation is less than the preset coefficient threshold, increase the number of sample points in the sample space to obtain a new sample space, repeat the above steps according to the new sample space to obtain a new neural network model, and calculate the failure rate and coefficient of variation until The calculated coding coefficient is greater than the preset coefficient threshold.

和上界

在最终生成失效率包括失效率上界和失效率下界，失效率即在失效率上界和实效率下界之间。Including the lower bound when calculating the failure rate of phononic crystals during service time

and upper bound

The final generation failure rate includes an upper bound of the failure rate and a lower bound of the failure rate, and the failure rate is between the upper bound of the failure rate and the lower bound of the actual rate.

The invention discloses a time-varying reliability testing method of a phononic crystal, which is suitable for execution in a computing device. The method includes the steps of: determining a sample space according to the parameters of the phononic crystal; building a reliability test model according to the sample space and the failure conditions of the phononic crystal; building a neural network model according to the reliability test model; predicting the phonon according to the neural network model and the sample space The failure rate of a crystal over its service time. By constructing the reliability test model of the phononic crystal and constructing the neural network model through the reliability test model, the invention can predict the failure rate of the phononic crystal within the service time, and avoid the inability to judge due to the lack of parameters of the phononic crystal. The performance of the phononic crystal realizes the determination of the performance of the phononic crystal during the service time.

In the description provided herein, numerous specific details are set forth. It will be understood, however, that embodiments of the invention may be practiced without these specific details. In some instances, well-known methods, structures and techniques have not been shown in detail in order not to obscure an understanding of this description.

A9. The method according to A8, wherein the determining the incremental sample point according to the number of iterations includes the steps of:

Determine the candidate point set according to the training set, and determine the first point set according to the candidate point set;

determining a second set of points from the first set of points by weight sampling;

Incremental sample points are determined from the second set of points according to an active learning function.

A10. The method according to A9, wherein the determining the incremental sample points from the second point set according to the active learning function comprises the steps of:

determining the uncertainty of each sample point in the second set of points;

determining the Euclidean distance between each sample point in the second point set and the training set;

Input the uncertainty and Euclidean distance of each sample point into the active learning function to obtain the function value of each sample point;

The sample point with the smallest function value in the second point set is taken as the incremental sample point.

A11. The method according to A10, wherein the determining the uncertainty of each sample point in the second point set includes the steps of:

Generate multiple training complements based on the last generated sample set;

Generate complementary neural network models from each training complement;

Determine the uncertainty of each sample point based on the supplementary neural network model and the last generated neural network model.

A12. The method according to any one of A1-A11, wherein the predicting the failure rate of the phononic crystal within the service time according to the neural network model and the sample space comprises the steps of:

The failure rate of the mixed time-varying reliability of the phononic crystal is calculated according to the neural network model and the sample points in the sample space.

A13. The method of A12, wherein the method further comprises the steps of:

Calculate the coefficient of variation of the failure rate according to the failure rate of the hybrid time-varying reliability of the phononic crystal;

determining whether the coefficient of variation is greater than a preset coefficient threshold;

If the coefficient of variation is not greater than the preset coefficient threshold, adding a new sample point to the sample space to obtain a new sample space;

The neural network model is trained according to the new sample space until the coefficient of variation of the neural network model is greater than the preset minimization threshold.

Similarly, it is to be understood that in the above description of exemplary embodiments of the invention, various features of the invention are sometimes grouped together into a single embodiment, figure, or its description.

Those skilled in the art will appreciate that modules or units or groups of devices in the examples disclosed herein may be arranged in the device as described in this embodiment, or alternatively may be positioned in the same way as the device in this example in one or more different devices. The modules in the preceding examples may be combined into one module or further divided into sub-modules.

Those skilled in the art will understand that the modules in the device in the embodiment can be adaptively changed and arranged in one or more devices different from the embodiment. The modules or units or inter-groups in the embodiments may be combined into one module or unit or inter-group, and further they may be divided into multiple sub-modules or sub-units or sub-groups. All features disclosed in this specification and all processes or elements of any method or apparatus so disclosed may be combined in any combination, except that at least some of such features and/or procedures or elements are mutually exclusive. Unless expressly stated otherwise, each feature disclosed in this specification may be replaced by alternative features serving the same, equivalent or similar purpose.

Furthermore, those skilled in the art will appreciate that although some of the embodiments described herein include certain features, but not others, included in other embodiments, that combinations of features of different embodiments are intended to be within the scope of the invention within and form different embodiments.

Furthermore, some of the described embodiments are described herein as methods or combinations of method elements that can be implemented by a processor of a computer system or by other means for performing the described functions. Thus, a processor having the necessary instructions for implementing the method or method element forms means for implementing the method or method element. Furthermore, an element of an apparatus embodiment described herein is an example of a means for carrying out the function performed by the element for the purpose of carrying out the invention.

The various techniques described herein can be implemented in conjunction with hardware or software, or a combination thereof. Thus, the method and apparatus of the present invention, or some aspects or portions of the method and apparatus of the present invention, may take the form of program code embedded in a tangible medium, such as a floppy disk, CD-ROM, hard drive, or any other machine-readable storage medium (ie, instructions), wherein when a program is loaded into a machine, such as a computer, and executed by the machine, the machine becomes an apparatus for practicing the invention.

Where the program code is executed on a programmable computer, the computing device typically includes a processor, a storage medium readable by the processor (including volatile and nonvolatile memory and/or storage elements), at least one input device, and at least one output device. Wherein, the memory is configured to store program codes; the processor is configured to execute the time-varying reliability testing method for phononic crystals of the present invention according to the instructions in the program codes stored in the memory.

By way of example and not limitation, computer-readable media includes computer storage media and communication media. Computer-readable media includes computer storage media and communication media. Computer storage media store information such as computer readable instructions, data structures, program modules or other data. Communication media typically embodies computer readable instructions, data structures, program modules or other data in a modulated data signal such as a carrier wave or other transport mechanism and includes any information delivery media. Combinations of any of the above are also included within the scope of computer-readable media.

As used herein, unless otherwise specified, the use of the ordinal numbers "first," "second," "third," etc. to describe common objects merely refers to different instances of similar objects, and is not intended to imply such The objects being described must have a given order in time, space, ordinal, or in any other way.

While the invention has been described in terms of a limited number of embodiments, those skilled in the art will appreciate, having the benefit of the above description, that other embodiments are conceivable within the scope of the invention thus described. Furthermore, it should be noted that the language used in this specification has been principally selected for readability and teaching purposes, rather than to explain or define the subject matter of the invention. Accordingly, many modifications and variations will be apparent to those skilled in the art. With regard to the scope of the present invention, the disclosure of the present invention is intended to be illustrative and not restrictive.

Claims (10)

1. A time-varying reliability testing method for phononic crystals, suitable for execution in a computing device, the method comprising the steps: Determine the sample space according to the parameters of the phononic crystal; constructing a reliability test model according to the sample space and the failure conditions of the phononic crystal; Build a neural network model according to the reliability test model; The failure rate of the phononic crystal in service time is predicted according to the neural network model and the sample space. 2. The method of claim 1, wherein the parameters of the phononic crystal comprise: random variable parameters, random process parameters, interval variable parameters and interval process parameters; Building a reliability test model according to the sample space and the failure conditions of the phononic crystal includes the steps: Convert the random process parameters in the crystal parameters to obtain equivalent random variable parameters; Convert the interval process parameters in the crystal parameters to obtain equivalent interval variable parameters; Convert the time parameter of the service time of the phononic crystal into the equivalent distribution time parameter; The reliability test model is constructed according to random variable parameters, equivalent random variable parameters, interval variable parameters, equivalent interval variable parameters, equivalent distribution time parameters and failure conditions. 3. The method of claim 2, wherein the failure condition comprises: During the service time, when the lower limit of the band gap of the phononic crystal is greater than the preset frequency, the phononic crystal fails. 4. The method according to any one of claims 1-3, wherein the building a neural network model according to the reliability test model comprises the steps: Determine the training set according to the sample point set; The neural network model is trained according to the reliability test model and the training set. 5. The method as claimed in claim, wherein, determining the training set according to the sample point set comprises the steps of: Converting the sample points in the sample space to obtain the converted sample point set; The training set is determined according to the transformed sample point set. 6. The method according to claim 5, wherein, determining the training set according to the transformed sample point set comprises the steps of: Determine its sample weight for each sample point in the transformed sample point set; Determine the characteristic value of each sample point according to the sample weight; The training set is determined from the transformed sample point set according to the feature value. 7. The method of claim 5, wherein the transforming the sample points in the sample space comprises the steps of: Equivalent uncertainty transformation is performed on each sample point in the sample space to obtain a time-independent sample point set. 8. The method of claim 4, wherein the method further comprises the step of: Judging whether the failure rate of the instantaneous reliability calculated by the neural network model satisfies the stopping rule; If the stopping rule is not satisfied, the number of iterations is set, and the incremental sample points are determined according to the number of iterations; Determine a new sample set according to the incremental sample points and the sample set; A new neural network model is trained according to the new sample set until a neural network model that satisfies the stopping rule is obtained. 9. A computing device comprising: one or more processors; memory; and One or more apparatuses comprising instructions for performing the method of any of claims 1-8. 10. A computer-readable storage medium storing one or more programs comprising instructions that, when executed by a computing device, cause the computing device to perform the functions according to claims 1-8 The method of any one.

Images