CN118889936A - An adaptive intelligent control method for explosion-proof motors - Google Patents

Abstract

The invention discloses a self-adaptive intelligent control method for an explosion-proof motor, which relates to the technical field of intelligent control, and comprises the following steps: and acquiring real-time working condition information of a target application scene, wherein the real-time working condition information comprises a real-time environment information set and a real-time intrinsic information set. Acquiring subdivision application scenes of the target explosion-proof motor, acquiring a historical risk event set by combining big data, and training a risk assessment model. And constructing a degradation objective function based on the risk assessment model to form a stepping degradation channel comprising a multi-layer cascade degradation sub-channel and a layer jump connection competition sub-channel. And initializing a stepping degradation channel based on the real-time environment information set and the real-time intrinsic information set. And acquiring the real-time working condition requirement of the target explosion-proof motor, and configuring a fluctuation weight sequence and a fluctuation constraint set. And activating a stepping degradation channel, and performing self-adaptive intelligent control by combining the fluctuation weight sequence and the fluctuation constraint set. Thereby achieving the technical effects of self-adaptive stable control and improving the operation safety before overhaul.

Description

An adaptive intelligent control method for explosion-proof motors

Technical Field

The invention relates to the field of intelligent control technology, and in particular to an adaptive intelligent control method for explosion-proof motors.

Background Art

Explosion-proof motors are widely used in explosive environments such as petroleum, chemical industry, and mining, and their safety is directly related to production and personal safety. Existing explosion-proof motor control methods mainly rely on fixed protection measures and preset working conditions, which makes it difficult to discover potential risks and lack intermediate transition control states. The reaction time reserved for maintenance and emergency disposal is short, and there are technical problems such as extensive control and unstable control transition that affect safety.

Summary of the invention

The present invention provides an adaptive intelligent control method for explosion-proof motors to solve the technical problems of extensive control and unstable control transition affecting safety in the prior art, achieve adaptive smooth control, and improve the technical effect of operating safety before maintenance.

The present invention provides an explosion-proof motor adaptive intelligent control method, comprising:

Interactive target application scenarios, collect and obtain real-time operating condition information, wherein the real-time operating condition information includes a real-time environmental information set and a real-time intrinsic information set; obtain segmented application scenarios of the target explosion-proof motor, obtain a historical risk event set based on the segmented application scenarios in combination with big data, and train and obtain a risk assessment model based on the historical risk event set; construct a degradation objective function based on the risk assessment model, and construct a step-down channel in combination with the risk assessment model and the degradation objective function, wherein the step-down channel includes a multi-layer cascaded degradation sub-channel and a plurality of competing sub-channels connected by jump layers; initialize the step-down channel based on the real-time environmental information set and the real-time intrinsic information set; obtain the real-time operating condition requirements of the target explosion-proof motor, configure the fluctuation weight sequence and the fluctuation constraint set; activate the step-down channel in combination with the fluctuation weight sequence and the fluctuation constraint set, and perform adaptive intelligent control.

In a feasible implementation, the interactive target application scenario collects and obtains real-time working condition information, including:

According to the explosion-proof motor knowledge graph, an objective environment variable set and a motor intrinsic variable set are defined; based on the objective environment variable set, a sensor network of a target application scenario is activated to collect the real-time environment information set; based on the motor intrinsic variable set, a target explosion-proof motor is accessed to obtain a real-time intrinsic information set.

In a feasible implementation, based on the risk assessment model, a degradation objective function is constructed, including:

Input the historical risk event set into the risk assessment model to obtain assessment output data; combine the assessment output data with the historical risk event set to construct an enhanced data set; based on the computing power characteristics and response requirements of the target application scenario, build a lightweight assessment model, and use the enhanced data set as training data to perform supervised training on the lightweight assessment model; define a degradation objective function based on the trained lightweight assessment model, wherein the degradation objective function is negatively correlated with the output of the lightweight assessment model.

In a feasible implementation, the risk assessment model and the degradation objective function are combined to construct a step-down channel, including:

Based on the backtracking optimization algorithm, the degradation objective function is used as the cost function to construct N layers of the degradation sub-channels, wherein the N layers of the degradation sub-channels are configured with N equally spaced cost function thresholds; based on the multi-objective optimization algorithm, the degradation objective function is used as the cost function to construct multiple competition sub-channels, wherein the competition sub-channels are configured with an interleaving step and a skip step; according to the skip step, a first optimization step of the degradation sub-channel and a second optimization step of the competition sub-channel are configured; and skip connections are established between the multiple competition sub-channels and the N layers of the degradation sub-channels to generate the step-by-step degradation channel.

In a feasible implementation, initializing the step-down channel based on the real-time environment information set and the real-time intrinsic information set includes:

Based on the real-time environment information set, the parameter value of the degradation objective function is configured; based on the real-time intrinsic information set, the initial control parameter set of the step-down channel is initialized, and the competition delay coefficients of the multiple competition sub-channels are defined.

In a feasible implementation, establishing a hopping connection between a plurality of the competing sub-channels and N layers of the degraded sub-channels to generate the step-by-step degraded channel, wherein the staggered hopping connections of the plurality of the competing sub-channels are connected to the multi-layer cascaded degraded sub-channels, comprises:

Establishing a unidirectional connection between an input end of a first competition sub-channel and an input end of a first degradation sub-channel, and establishing a unidirectional connection between an output end of the first competition sub-channel and an input end of an ath degradation sub-channel based on a preset layer hopping step a, wherein a is a positive integer and a is less than or equal to N;

Based on the interleaving step length b, a unidirectional connection is established between the input end of the second competition subchannel and the input end of the a-bth degradation subchannel, and a unidirectional connection is established between the output end of the second competition subchannel and the input end of the 2a-bth degradation subchannel, wherein b is a positive integer and b is less than a;

Traversing N layers of the degradation sub-channels, k of the competition sub-channels are configured, wherein the difference in the number of layers between the output end of the kth competition sub-channel and the output end of the Nth degradation sub-channel is less than a.

In a feasible implementation, the fluctuation weight sequence and the fluctuation constraint set are combined to activate the step-down channel and perform adaptive intelligent control, including:

According to the initial control parameter set, the first degraded sub-channel is activated for iterative parameter optimization; when the iterative progress of the first degraded sub-channel meets the competition delay coefficient, the real-time control parameter set is synchronized to the first competition sub-channel for iterative competition optimization; the first degraded sub-channel transmits the parameter optimization result to the second degraded sub-channel, and performs step-by-step optimization based on the N layers of the degraded sub-channel until the ath degraded sub-channel.

If the first competition sub-channel reaches the ath cost function threshold before the ath degradation sub-channel, the parameter optimization result of the first competition sub-channel is transmitted to the a+1th degradation sub-channel, and the parameter optimization result of the a-bth degradation sub-channel is transmitted to the second competition sub-channel for iterative competition optimization.

Traversing the N layers of the degraded sub-channels, and performing multiple iterations of competitive optimization with the k competing sub-channels, until the parameter optimization result of any layer of the degraded sub-channels or any one of the competing sub-channels meets the risk control threshold of the target scenario.

In a feasible implementation, if the ath degraded sub-channel reaches the ath cost function threshold before the first competitive sub-channel, the parameter optimization result of the ath degraded sub-channel is transmitted to the a+1th degraded sub-channel, and the parameter optimization result of the a-bth degraded sub-channel is transmitted to the second competitive sub-channel for iterative competitive optimization.

The present invention discloses an adaptive intelligent control method for explosion-proof motors, including: interactive target application scenarios, collecting real-time operating information, including real-time environmental information sets and real-time intrinsic information sets. Obtaining segmented application scenarios of target explosion-proof motors, combining big data to obtain historical risk event sets, and training risk assessment models. Based on the risk assessment model, a degradation objective function is constructed, and a step-down channel is constructed in combination with the degradation objective function, which includes multi-layer cascaded degradation sub-channels and skip-layer connected competition sub-channels. The step-down channel is initialized using the real-time environmental information set and the real-time intrinsic information set. Obtaining the real-time operating requirements of the explosion-proof motor, configuring the fluctuation weight sequence and the fluctuation constraint set. In combination with the fluctuation weight sequence and the fluctuation constraint set, the step-down channel is activated to perform adaptive intelligent control. The adaptive intelligent control method for explosion-proof motors disclosed by the present invention solves the technical problems of extensive control and unstable control transition affecting safety, realizes adaptive smooth control, and improves the technical effect of operating safety before maintenance.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG1 is a schematic flow chart of an adaptive intelligent control method for explosion-proof motors according to the present invention.

FIG. 2 is a schematic diagram of a flow chart of constructing a step-down channel in an adaptive intelligent control method for explosion-proof motors according to the present invention.

DETAILED DESCRIPTION

The technical solution provided in the embodiments of the present invention is to solve the technical problems of extensive control and unstable control transition affecting safety in the prior art. The overall idea adopted is as follows:

First, the target application scenario is interacted with to collect and obtain real-time operating condition information, wherein the real-time operating condition information includes a real-time environmental information set and a real-time intrinsic information set. Then, the subdivided application scenario of the target explosion-proof motor is obtained, and based on the subdivided application scenario, a historical risk event set is obtained in combination with big data, and a risk assessment model is trained and obtained based on the historical risk event set. Then, based on the risk assessment model, a degradation objective function is constructed, and a step degradation channel is constructed in combination with the risk assessment model and the degradation objective function, wherein the step degradation channel includes a multi-layer cascaded degradation sub-channel and a plurality of competitive sub-channels connected by a jump layer. Next, based on the real-time environmental information set and the real-time intrinsic information set, the step degradation channel is initialized. Then, the real-time operating condition requirements of the target explosion-proof motor are obtained, and the fluctuation weight sequence and the fluctuation constraint set are configured. Finally, the step degradation channel is activated in combination with the fluctuation weight sequence and the fluctuation constraint set to perform adaptive intelligent control.

The above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods of the specification to better understand the above technical solution. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments of the present invention. It should be understood that the present invention is not limited to the example embodiments used only to explain the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. In addition, it should be noted that, for the convenience of description, only the parts related to the present invention are shown in the drawings, rather than all of them.

Embodiment 1

FIG1 is a flow chart of an adaptive intelligent control method for explosion-proof motors according to the present invention, wherein the method comprises:

In the interactive target application scenario, real-time operating condition information is collected and acquired, wherein the real-time operating condition information includes a real-time environment information set and a real-time intrinsic information set.

In some embodiments, the interactive target application scenario and the acquisition of real-time working condition information include:

According to the explosion-proof motor knowledge graph, an objective environment variable set and a motor intrinsic variable set are defined; based on the objective environment variable set, a sensor network of a target application scenario is activated to collect the real-time environment information set; based on the motor intrinsic variable set, a target explosion-proof motor is accessed to obtain a real-time intrinsic information set.

Specifically, firstly, based on the existing explosion-proof motor knowledge graph, a set of relevant variables related to the operation risks of explosion-proof motors is defined, including the objective environment variable set and the motor itself variable set.

Among them, the objective environment variable set refers to various factors in the motor operating environment, which may affect the operating state of the motor. For example, it includes ambient temperature, ambient humidity, ambient pressure, vibration, dust concentration, corrosive level, etc. The motor self variable set refers to various parameters and states of the motor itself. The indicators and parameters of the motor self variable set help to understand the operating conditions of the motor. For example, it includes voltage, current, speed, torque (torque), shaft power, efficiency, temperature (winding temperature), vibration, etc.

Specifically, according to the defined set of environmental variables, the corresponding sensor network is activated to collect real-time environmental information to understand the environmental conditions of the motor operation. The sensor network is composed of a series of sensors, each of which is responsible for monitoring one or more specific parameters. Exemplary sensors include temperature sensors, humidity sensors, pressure sensors, vibration sensors, dust sensors, etc.

Specifically, according to the defined motor self variable set, the target explosion-proof motor is accessed to obtain the real-time intrinsic information of the target explosion-proof motor. First, it is connected to the control component or system of the motor, such as PLC (Programmable Logic Controller) or SCADA (Supervisory Control and Data Acquisition) system, through wired or wireless means. Then, based on the preset transmission interface and transmission protocol, the speed, load, current, voltage, temperature and other information of the target explosion-proof motor are collected, and the information is verified and cleaned, and the information unit is standardized and stored as real-time intrinsic information.

Through the above steps, the operating status and environmental status of the explosion-proof motor can be monitored in real time, providing a data basis for subsequent risk analysis and control decisions, and facilitating timely warning and handling of potential risks.

Obtain segmented application scenarios of the target explosion-proof motor, obtain a set of historical risk events based on the segmented application scenarios in combination with big data, and train a risk assessment model based on the set of historical risk events.

Specifically, first, identify and define specific application scenarios of explosion-proof motors, such as chemical plants, coal mines, oil platforms, etc., and determine them as segmented application scenarios; then, use big data platforms and technologies to determine multi-source data sets containing historical risk events, such as equipment failure records, safety accident reports, environmental monitoring data, etc., and collect historical risk event data from various data sources to generate a historical risk event set. Among them, the historical risk event set includes past motor failure situations, historical operating conditions information, and failure rates of various types of failures.

Specifically, to train and acquire a risk assessment model, first, clean the collected historical risk event set to remove noise and invalid data. Then, annotate the cleaned data to indicate the event type, cause of occurrence (i.e., strongly associated working condition information), degree of impact, and other information. Then, select a suitable machine learning model, such as decision tree, random forest, neural network, etc. Using the annotated historical risk event data set as training data, supervised training is performed on the selected model to learn and acquire the ability to assess risk levels based on working condition information, and generate a training risk assessment model.

Through the above method, historical fault and accident data are collected, and then a risk assessment model is trained to evaluate the operating risks of explosion-proof motors in real time and provide early warning and decision support, which helps to improve production safety.

Based on the risk assessment model, a degradation objective function is constructed, and the risk assessment model and the degradation objective function are combined to construct a step-down channel, wherein the step-down channel includes multiple layers of cascaded degradation sub-channels and multiple competing sub-channels connected by skip layers.

Specifically, the step-down channel includes multiple downgrade sub-channels and multiple competing sub-channels that are arranged in parallel and connected by skipping layers, wherein the multiple downgrade sub-channels are connected step by step, that is, the output end of the first-order downgrade sub-channel is connected to the input end of the next-order downgrade sub-channel. Multiple competing sub-channels are skip-connected between the multiple downgrade sub-channels.

Specifically, the multi-layer cascaded degradation sub-channel is used for fine parameter optimization and selection of degradation paths, which is used to gradually approach the required degraded control parameters with a smaller parameter adjustment step size, and the multiple competing sub-channels have a larger parameter adjustment step size, which is used to try to quickly approach the required degraded control parameters, and the degradation targets of the multiple competing sub-channels are deeper than the degradation targets of each degraded sub-channel. In other words, a single competing sub-channel is expected to achieve the degradation effect of multiple degraded sub-channels.

For example, if the degradation target of each degradation sub-channel is to improve the safety of the control parameter by 2%, the degradation target of the contention sub-channel may be any value that is greater than or equal to 2% in safety improvement.

The above-mentioned multi-layer cascaded degradation sub-channel and competition sub-channel design ensures that the step-down channel can take corresponding protection measures at different risk levels, thereby improving the security and reliability of the target application scenario.

In some embodiments, as shown in FIG2 , based on the risk assessment model, a degradation objective function is constructed, including:

Input the historical risk event set into the risk assessment model to obtain assessment output data; combine the assessment output data with the historical risk event set to construct an enhanced data set; based on the computing power characteristics and response requirements of the target application scenario, build a lightweight assessment model, and use the enhanced data set as training data to perform supervised training on the lightweight assessment model; define a degradation objective function based on the trained lightweight assessment model, wherein the degradation objective function is negatively correlated with the output of the lightweight assessment model.

Specifically, first, the historical risk event set is input into the risk assessment model to obtain the assessment output data, which is positive and negative labels marked with probabilities, and has richer information than the historical risk event set. In other words, the historical risk event set can be regarded as a Hard-target, and each event is clearly marked as a positive sample (correct risk assessment value) or a negative sample (wrong risk assessment value). Correspondingly, the assessment output data can be regarded as a Soft-target, which includes the probability distribution of the risk assessment model output, which is used to represent the uncertainty of the risk assessment model for each possible risk assessment value. These Soft-targets provide richer information than Hard-targets, which is conducive to making the update direction of the model more stable in subsequent training, thereby reducing gradient fluctuations and improving training efficiency.

Specifically, the output data of the risk assessment model is combined with the historical risk event set to create an enhanced dataset. The enhanced dataset contains both the Hard-target represented by the historical risk event set and the Soft-target represented by the assessment output data, which can provide more comprehensive features and help improve the accuracy and robustness of the lightweight assessment model.

Specifically, according to the computing power and response time requirements of the target application scenario, a lightweight assessment model is designed and trained using an enhanced data set so that the lightweight assessment model can learn the prediction mode and prediction results of the risk assessment model. This lightweight assessment model is simpler than the original assessment model and can still have high prediction performance in scenarios with limited computing power.

Specifically, a degradation objective function is defined based on the trained lightweight assessment model. The degradation objective function should be negatively correlated with the output of the lightweight assessment model, that is, the higher the risk assessment value, the smaller the degradation objective function value. By maximizing the degradation objective function, the performance degradation operation of the target explosion-proof motor can be guided under high-risk conditions, thereby improving the production safety of the target explosion-proof motor and the target application scenario.

In some embodiments, combining the risk assessment model with the degradation objective function to construct a step-down channel includes:

Based on the backtracking optimization algorithm, the degradation objective function is used as the cost function to construct N layers of the degradation sub-channels, wherein the N layers of the degradation sub-channels are configured with N equally spaced cost function thresholds; based on the multi-objective optimization algorithm, the degradation objective function is used as the cost function to construct multiple competition sub-channels, wherein the competition sub-channels are configured with an interleaving step and a skip step; according to the skip step, a first optimization step of the degradation sub-channel and a second optimization step of the competition sub-channel are configured; and skip connections are established between the multiple competition sub-channels and the N layers of the degradation sub-channels to generate the step-by-step degradation channel.

Specifically, a backtracking optimization algorithm is used, and the degradation target function is used as the cost function to construct N layers of degradation sub-channels layer by layer, wherein each layer of the N layers of degradation sub-channels is used to process the degradation sub-segments in the degradation target. In other words, each layer of degradation sub-channels has its own optimization target. Through the above method, a complex degradation problem can be decomposed into multiple simple sub-problems, and each sub-problem can be solved by a layer of degradation sub-channels.

Optionally, in N layers of degraded sub-channels, each degraded sub-channel needs to achieve the same degraded effect, that is, each degraded sub-channel needs to improve the same safety level of the control parameters, which helps to better balance the burden of each sub-channel and avoid the burden or oversimplification of some sub-channels.

Optionally, each downgraded sub-channel has the same model parameters, that is, the same model structure and parameter initialization method can be used to construct each sub-channel. This can effectively simplify the design and implementation process of the model, and is conducive to the training and optimization of the model. In addition, although the model parameters of each downgraded sub-channel are the same during initialization, during the training process, since each sub-channel needs to process different tasks, the defined parameter output constraints are different, that is, the task objectives of each downgraded sub-channel are different.

Specifically, in each layer of the degradation sub-channel, N equally spaced cost function thresholds are configured. The cost function threshold determines the triggering condition for outputting the degradation result of each layer.

Specifically, a multi-objective optimization algorithm (such as particle swarm optimization, genetic algorithm, imperial competition algorithm, etc.) is used to optimize the configuration of the competition sub-channel by taking the degradation objective function as the cost function. The competition sub-channel allows skipping some levels and directly entering a higher level of degradation measures.

Among them, the competition sub-channel is configured with an interleaving step and a skipping step to adapt to the changes of different risk levels. The skipping step specifies the cost function threshold or the number of layers of the downgraded sub-channel that the competition sub-channel can skip. The interleaving step determines whether there is overlap and the size of overlap between multiple competition sub-channels, that is, whether the subsequent competition sub-channel takes the downgraded sub-channel that has been skipped by the previous competition sub-channel as the starting point for downgrade optimization. The interleaving step and the skipping step determine the downgrade path and speed of the competition sub-channel, so that it can respond quickly to high-risk situations.

Specifically, the first optimization step size determines the optimization step size of each layer in the degradation sub-channel, so that it optimizes the degradation path layer by layer. The second optimization step size determines the optimization step size of each step in the competition sub-channel, so that it can quickly adapt to high-risk situations.

Specifically, the first optimization step length is smaller than the second optimization step length, thereby ensuring that the competition sub-channel has a faster convergence speed. Correspondingly, the optimization accuracy of the competition sub-channel is low, and the control parameters are prone to oscillation. At this time, the downgraded sub-channel based on the smaller first optimization step length can ensure the optimization accuracy of the control parameters of the downgrade process and ensure that the downgrade operation can be effectively completed.

Configuring the step-downgrade channel through the above steps and methods helps to improve the efficiency of the downgrade operation while ensuring the basic downgrade effect, thereby helping to improve the response performance of the adaptive control of the present application.

In some implementations, establishing a hop connection between the plurality of competing sub-channels and the N layers of the degradation sub-channels to generate the step-down channel, wherein the staggered hop connections of the plurality of competing sub-channels are connected to the multi-layer cascaded degradation sub-channels, comprises:

Establish a unidirectional connection between the input end of the first competition sub-channel and the input end of the first degradation sub-channel, and establish a unidirectional connection between the output end of the first competition sub-channel and the input end of the a-th degradation sub-channel based on a preset jump step a, wherein a is a positive integer and a is less than or equal to N; establish a unidirectional connection between the input end of the second competition sub-channel and the input end of the a-b-th degradation sub-channel based on an interleaving step b, and establish a unidirectional connection between the output end of the second competition sub-channel and the input end of the 2a-b-th degradation sub-channel, wherein b is a positive integer and b is less than a; traverse N layers of the degradation sub-channels and configure k competition sub-channels, wherein the difference in the number of layers between the output end of the k-th competition sub-channel and the output end of the N-th degradation sub-channel is less than a.

Specifically, first, a unidirectional connection is established between the input end of the first competition subchannel and the input end of the first degradation subchannel, and based on the preset layer jump step a, a unidirectional connection is established between the output end of the first competition subchannel and the input end of the ath degradation subchannel. Thus, the first layer jump connection between the competition subchannel and the N-layer degradation subchannel is established. The above-mentioned unidirectional connection is used to control the flow direction of the initial input of the degradation optimization.

Specifically, based on the interleaving step length b, a unidirectional connection is established between the input end of the second competition subchannel and the input end of the a-bth degradation subchannel, and a unidirectional connection is established between the output end of the second competition subchannel and the input end of the 2a-bth degradation subchannel. The above steps are used to establish a second layer-hopping connection between the competition subchannel and the N-layer degradation subchannel. Among them, the second competition subchannel and the first competition subchannel both skip the b-layer degradation subchannel from the a-bth degradation subchannel to the a-th degradation subchannel.

Further, N layers of degraded sub-channels are traversed to configure k competing sub-channels, wherein the difference in the number of layers between the output end of the k-th competing sub-channel and the output end of the N-th degraded sub-channel is less than a. In other words, when the number of layers of the remaining degraded sub-channels that have not been skipped is less than the skip step a, it is considered that all skip-layer connections between the competing sub-channels and the degraded sub-channels have been established.

The above method helps the step-down channel to transmit information more effectively during the optimization process and improve the performance of the network. Specifically, the competitive subchannel can quickly find a better solution, and then pass this solution to the subsequent downgraded subchannel through the skip layer connection. The downgraded subchannel is used for fine optimization to find a more accurate optimal solution. By staggering the step size, the control parameters generated by the intermediate downgrade process of the downgraded subchannel can be retained, which helps to reduce the risk of overfitting.

Furthermore, the step-downgrade channel combines layer-by-layer optimization with competitive optimization to achieve efficient and flexible downgrade measures, ensuring that appropriate downgrade operations can be taken at different risk levels.

The step-down channel is initialized based on the real-time environment information set and the real-time intrinsic information set.

In some embodiments, initializing the step-down channel based on the real-time environment information set and the real-time intrinsic information set includes:

Based on the real-time environment information set, the parameter value of the degradation objective function is configured; based on the real-time intrinsic information set, the initial control parameter set of the step-down channel is initialized, and the competition delay coefficients of the multiple competition sub-channels are defined.

Specifically, according to the extracted environmental parameters, the parameter value of the degradation objective function is set so that it can reflect the actual situation of the current environment. For example, if the temperature is too high, the parameter value of the degradation objective function should reflect the need for rapid degradation.

Specifically, according to the extracted motor state parameters, a multi-objective optimization algorithm such as a particle swarm algorithm (PSO) or a genetic algorithm (GA) is used to perform random fluctuations to obtain the initial control parameter set of the step-down channel. The initial control parameter set is the basic control parameter when starting the downgrade, ensuring that the downgrade process starts from a reasonable state. Optionally, according to the specific needs of the target application scenario, set relevant parameters such as the size of the initial control parameter set, the number of iterations, the mutation rate, etc.

Specifically, according to the target application scenario and historical data, a competition delay coefficient is set, which defines the timing of the competition sub-channel synchronizing the control parameter set from the downgraded sub-channel, and is used to ensure that the competition sub-channel can synchronize parameters at the optimal time, thereby obtaining the initial optimization direction and improving the efficiency of the downgrade operation. Exemplarily, the competition delay coefficient is set to the ratio of the number of iterations to the number of planned iterations of the downgraded sub-channel, that is, the iterative progress of the control parameter downgrade operation.

Through the above method steps, the step-down channel can be helped to better adapt to the control needs of the target application scenario, and at the same time, a better solution and optimization direction can be found more quickly for the competing sub-channel, thereby improving the efficiency of the downgrade operation.

Obtain the real-time operating requirements of the target explosion-proof motor and configure the fluctuation weight sequence and fluctuation constraint set.

Specifically, the fluctuation weight sequence is used to specify the sequence of control parameter adjustment priorities. Parameters with higher weights will be adjusted first. For example, for explosion-proof motors used in explosion-proof elevators, key parameters such as torque and temperature (affecting continuous working ability) are given lower priority. These parameters are crucial to the safety and stability of the elevator, so when adjusting the parameters, these parameters should be kept at a high level as much as possible. Correspondingly, non-critical parameters such as noise, life, speed, speed constant, torque constant, motor efficiency, etc. have less impact on the basic functions and safety of the elevator. They are given higher priority for adjustment.

Optionally, when adjusting the weights of these non-critical parameters, the possible impact on critical parameters also needs to be considered. For example, excessively increasing the speed may cause a decrease in torque to ensure overall performance and safety.

Specifically, the fluctuation constraint set is used to ensure the safety lower limit of the key performance parameters of the motor during operation. For example, for explosion-proof motors used in explosion-proof elevators, the minimum safety value of each key performance parameter is set according to the application requirements of the explosion-proof elevator and stored as a fluctuation constraint set. Including: torque lower limit, to ensure that the elevator can operate normally and avoid risks safely, function continuous capacity lower limit, to prevent motor overheating and functional failure, etc.

The fluctuation weight sequence is combined with the fluctuation constraint set, the step-down channel is activated, and adaptive intelligent control is performed.

In some embodiments, combining the fluctuation weight sequence with the fluctuation constraint set, activating the step-down channel, and performing adaptive intelligent control include:

According to the initial control parameter set, the first degraded sub-channel is activated for iterative parameter optimization; when the iterative progress of the first degraded sub-channel meets the competition delay coefficient, the real-time control parameter set is synchronized to the first competition sub-channel for iterative competition optimization; the first degraded sub-channel transmits the parameter optimization result to the second degraded sub-channel, and performs step-by-step optimization based on the N layers of degraded sub-channels until the ath degraded sub-channel; if the first competition sub-channel reaches the ath cost function threshold before the ath degraded sub-channel, the parameter optimization result of the first competition sub-channel is transmitted to the a+1th degraded sub-channel, and the parameter optimization result of the a-bth degraded sub-channel is transmitted to the second competition sub-channel for iterative competition optimization; traverse the N layers of the degraded sub-channels, and perform multiple iterative competition optimizations with the k competition sub-channels until the parameter optimization result of any layer of the degraded sub-channels or any one of the competition sub-channels meets the risk control threshold of the target scenario.

Specifically, first, the initial control parameter set is updated by combining the fluctuation weight sequence and the fluctuation constraint set, and the multi-layer degradation sub-channel is input to start the degradation optimization iteration. When the iteration progress of the first degradation sub-channel meets the competition delay coefficient, the real-time control parameter set is synchronized to the first competition sub-channel to start the iterative competition optimization. And the degradation optimization iteration of the first degradation sub-channel is continued.

Specifically, after the first downgrade sub-channel completes the iteration, the parameter optimization result is transmitted to the second downgrade sub-channel, and multiple layers of downgrade sub-channels are activated in sequence to perform downgrade optimization iterations, parameter optimization is performed at each layer, and control parameters are adjusted to gradually approach the target optimization result to achieve a step-by-step downgrade operation until the step-by-step downgrade reaches the ath downgrade sub-channel.

Specifically, when the first a downgraded subchannels and the first competitive subchannel perform synchronous step downgrade and iterative competitive optimization, the order of the first a downgraded subchannels and the first competitive subchannel in achieving the ath cost function threshold corresponding to the ath downgraded subchannel is determined. If the first competitive subchannel reaches the ath cost function threshold before the ath downgraded subchannel, the step downgrade of the first a downgraded subchannels is stopped, the optimization result of the first competitive subchannel is used as the input of the a+1th downgraded subchannel, and the optimization result of the a-bth downgraded subchannel is transmitted to the second competitive subchannel.

Optionally, if the step-downgrade of the first a downgraded sub-channels has not yet been transmitted to the a-bth downgraded sub-channel, the real-time intermediate optimization results of the first a downgraded sub-channels are used as the optimization result of the a-bth downgraded sub-channel as the input of the second competitive sub-channel.

Specifically, all N layers of downgraded sub-channels and k competitive sub-channels are traversed, and multiple iterations of competitive optimization are performed to ensure that the control parameter adjustment meets the risk control threshold of the target scenario. When the parameter optimization result of any layer of downgraded sub-channel or any competitive sub-channel meets the risk control threshold of the target scenario, the iteration is stopped and the final optimized control parameter set is output for intelligent control in practical applications. Among them, the risk control threshold is set according to specific business needs, safety standards and other factors, and represents the maximum risk level allowed in a specific scenario (target application scenario).

In some embodiments, the method further comprises:

If the ath degraded sub-channel reaches the ath cost function threshold before the first competitive sub-channel, the parameter optimization result of the ath degraded sub-channel is transmitted to the a+1th degraded sub-channel, and the parameter optimization result of the a-bth degraded sub-channel is transmitted to the second competitive sub-channel for iterative competitive optimization.

Furthermore, if the a-th degradation subchannel reaches the a-th cost function threshold before the first competition subchannel, the parameter optimization result of the a-th degradation subchannel is transmitted to the a+1-th degradation subchannel. The parameter optimization result of the a-b-th degradation subchannel is transmitted to the second competition subchannel for iterative competition optimization. In other words, the order in which the a-th degradation subchannel and the first competition subchannel reach the a-th cost function threshold determines which parameter optimization result of the two is retained as the input of the subsequent degradation subchannel.

Specifically, because the optimization convergence efficiency of multiple degraded sub-channels is relatively low, priority is given to ensuring the continuity of the step-by-step degradation of multiple degraded sub-channels, and using the parameter optimization results that first meet the cost function threshold as the input of subsequent degraded sub-channels. At the same time, the intermediate optimization results of the degraded sub-channels are retained, and synchronous degradation operations are performed through competitive sub-channels with higher convergence efficiency. As many feasible degradation optimization methods as possible are retained to improve the stability of adaptive intelligent control.

In summary, the explosion-proof motor adaptive intelligent control method provided by the present invention has the following technical effects:

Through interactive target application scenarios, real-time operating information is collected, including real-time environmental information sets and real-time intrinsic information sets. Obtain the segmented application scenarios of the target explosion-proof motor, combine big data to obtain the historical risk event set, and train the risk assessment model. Based on the risk assessment model, a degradation objective function is constructed, and a step degradation channel is constructed in combination with the degradation objective function. The channel includes multi-layer cascaded degradation sub-channels and skip-layer connected competition sub-channels. Use the real-time environmental information set and the real-time intrinsic information set to initialize the step degradation channel. Obtain the real-time operating condition requirements of the explosion-proof motor, configure the fluctuation weight sequence and fluctuation constraint set. Combine the fluctuation weight sequence and fluctuation constraint set to activate the step degradation channel and perform adaptive intelligent control. Then achieve adaptive smooth control and improve the technical effect of operating safety before maintenance.

It should be understood that the embodiments disclosed in the present invention and the above description can enable those skilled in the art to use the present invention to implement the present invention. At the same time, the present invention is not limited to the above-mentioned embodiments. It should be understood that those skilled in the art can still modify the technical solutions recorded in the above-mentioned embodiments, or replace some of the technical features therein by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.

Claims (8)

1. An adaptive intelligent control method for an explosion-proof motor is characterized by comprising the following steps:

The method comprises the steps of interacting a target application scene, and acquiring real-time working condition information, wherein the real-time working condition information comprises a real-time environment information set and a real-time intrinsic information set;

acquiring subdivision application scenes of a target explosion-proof motor, acquiring a historical risk event set based on the subdivision application scenes and combining big data, and training and acquiring a risk assessment model based on the historical risk event set;

constructing a degradation objective function based on the risk assessment model, and combining the risk assessment model and the degradation objective function to construct a stepping degradation channel, wherein the stepping degradation channel comprises a plurality of cascade degradation sub-channels and a plurality of competition sub-channels connected with a layer jump;

initializing the step degradation channel based on the real-time environmental information set and the real-time intrinsic information set;

Acquiring the real-time working condition requirement of a target explosion-proof motor, and configuring a fluctuation weight sequence and a fluctuation constraint set;

And activating the stepping degradation channel by combining the fluctuation weight sequence and the fluctuation constraint set, and performing self-adaptive intelligent control.

2. The method for adaptively and intelligently controlling an explosion-proof motor according to claim 1, wherein the steps of interacting a target application scene, acquiring real-time working condition information, and the method comprises the following steps:

according to the explosion-proof motor knowledge graph, defining an objective environment variable set and a motor self variable set;

Activating a sensor network of a target application scene based on the objective environment variable set, and collecting the real-time environment information set;

based on the motor self variable set, accessing the target explosion-proof motor to obtain a real-time intrinsic information set.

3. The adaptive intelligent control method of an explosion-proof motor according to claim 2, wherein constructing a degradation objective function based on the risk assessment model comprises:

inputting the historical risk event set to the risk assessment model to obtain assessment output data;

combining the evaluation output data with the historical risk event set to construct an enhanced data set;

Constructing a lightweight evaluation model based on the computational power characteristics and response requirements of a target application scene, and performing supervision training on the lightweight evaluation model by taking the enhanced data set as training data and taking the enhanced data set as a training data set;

A degradation objective function is defined based on the lightweight evaluation model that is trained, wherein the degradation objective function is inversely related to an output of the lightweight evaluation model.

4. The method of claim 3, wherein constructing a stepped degradation channel by combining the risk assessment model and the degradation objective function comprises:

Based on a backtracking optimization algorithm, constructing N layers of degradation sub-channels by taking the degradation objective function as a cost function, wherein N equally-spaced cost function thresholds are configured for the N layers of degradation sub-channels;

Based on a multi-objective optimization algorithm, constructing a plurality of competing sub-channels by taking the degradation objective function as a cost function, wherein the competing sub-channels are configured with staggered step sizes and layer jump step sizes;

According to the layer jump step length, configuring a first optimizing step length of the degradation sub-channel and a second optimizing step length of the competition sub-channel;

and establishing layer-jump connection of the plurality of competing sub-channels and the N layers of degradation sub-channels to generate the stepping degradation channel.

5. The method of claim 4, wherein initializing the step degradation channel based on the real-time environmental information set and the real-time intrinsic information set comprises:

configuring a parameter value of the degradation objective function based on the real-time environment information set;

And initializing an initial control parameter set of a stepping degradation channel based on the real-time intrinsic information set, and defining competition delay coefficients of a plurality of competition sub-channels.

6. The method of claim 5, wherein establishing a plurality of said competing sub-channels with the layer-hopping connections of N layers of said degrading sub-channels to generate said stepped degrading sub-channels, the staggered layer-hopping connections of the plurality of said competing sub-channels to the multi-layer cascade of said degrading sub-channels comprises:

establishing unidirectional connection between the input end of a first competitive sub-channel and the input end of a first degradation sub-channel, and establishing unidirectional connection between the output end of the first competitive sub-channel and the input end of an a degradation sub-channel based on a preset layer jump step length a, wherein a is a positive integer, and a is less than or equal to N;

based on the staggered step length b, establishing unidirectional connection between the input end of the second competition sub-channel and the input end of the a-b degradation sub-channel, and establishing unidirectional connection between the output end of the second competition sub-channel and the input end of the 2a-b degradation sub-channel, wherein b is a positive integer, and b is smaller than a;

traversing the degradation sub-channels of the N layers, and configuring k competition sub-channels, wherein the layer number difference between the output end of the kth competition sub-channel and the output end of the Nth degradation sub-channel is smaller than a.

7. The method of claim 6, wherein the step degradation channel is activated in combination with the fluctuation weight sequence and the fluctuation constraint set to perform the adaptive intelligent control, and the method comprises:

activating the first degradation sub-channel to perform iterative parameter optimization according to the initial control parameter set;

when the iteration progress of the first degradation sub-channel meets the competition delay coefficient, synchronizing a real-time control parameter set to the first competition sub-channel to perform iterative competition optimization;

The first degradation sub-channel transmits a parameter optimization result to a second degradation sub-channel, and the degradation sub-channels are optimized step by step based on N layers until an a degradation sub-channel is reached;

If the first competition sub-channel reaches an a cost function threshold value before the a degradation sub-channel, transmitting a parameter optimization result of the first competition sub-channel to the a+1 degradation sub-channel, and transmitting a parameter optimization result of the a-b degradation sub-channel to a second competition sub-channel to perform iterative competition optimization;

And traversing the degradation sub-channels of the N layers, and carrying out repeated iterative competition optimization with k competition sub-channels until the parameter optimization result of any layer of degradation sub-channels or any layer of competition sub-channels meets the risk control threshold of the target scene.

8. The adaptive intelligent control method for an explosion-proof motor according to claim 7, further comprising:

If the a-th degradation sub-channel reaches the a-th cost function threshold before the first competition sub-channel, transmitting a parameter optimization result of the a-th degradation sub-channel to the a+1-th degradation sub-channel, and transmitting a parameter optimization result of the a-b-th degradation sub-channel to a second competition sub-channel to perform iterative competition optimization.