Thermal shock testing machine control method, device, equipment and storage medium

Abstract

The present application relates to the field of testing machine control technology, and discloses a control method, device, equipment and storage medium for a hot and cold shock testing machine. The method includes: obtaining the physical structure and material properties of the hot and cold shock testing machine and constructing a temperature prediction model; arranging multiple temperature sensors in the test area and obtaining multiple temperature sensor data and performing adaptive weighted fusion to generate three-dimensional temperature distribution data of the test area; constructing a model prediction controller based on the temperature prediction model and the three-dimensional temperature distribution data, and solving the optimal control problem through the model prediction controller to generate a first temperature control sequence; obtaining the operating status data of the refrigeration system and establishing a corresponding thermodynamic model, and generating an adaptive refrigeration strategy for the refrigeration system based on the thermodynamic model; controlling and optimizing the first temperature control sequence based on the adaptive refrigeration strategy to generate a second temperature control sequence. The present application improves the control accuracy of the hot and cold shock testing machine.

Description

Technical Field

[0001] The present application relates to the field of testing machine control technology, and in particular to a control method, device, equipment and storage medium for a thermal shock testing machine. Background Art

[0002] Thermal shock testers are used to evaluate the reliability of materials and products under extreme temperature changes. However, traditional thermal shock tester control methods have certain limitations. Due to the lack of accurate temperature prediction models, it is difficult to accurately simulate complex physical structures and material properties, resulting in insufficient temperature control accuracy.

[0003] The temperature control of the hot and cold shock test not only needs to respond quickly to temperature changes, but also needs to consider the energy efficiency ratio and operating stability of the refrigeration system. However, the existing control methods do not study the operating status and thermodynamic characteristics of the refrigeration system enough, and fail to form an adaptive refrigeration strategy, which restricts the comprehensive performance optimization of the test machine. In addition, most traditional temperature control sequence optimization methods rely on simple empirical algorithms, which are difficult to achieve global optimal control under complex working conditions, resulting in unsatisfactory temperature control accuracy and energy consumption optimization effects. Summary of the invention

[0004] The present application provides a control method, device, equipment and storage medium for a thermal shock testing machine, and the present application improves the control accuracy of the thermal shock testing machine.

[0005] In a first aspect, the present application provides a method for controlling a thermal shock tester, the method comprising:

[0006] Acquire the physical structure and material properties of the thermal shock testing machine, and construct a temperature prediction model of the thermal shock testing machine;

[0007] Arranging a plurality of temperature sensors in the test area and acquiring a plurality of temperature sensor data, and performing adaptive weighted fusion on the plurality of temperature sensor data to generate three-dimensional temperature distribution data of the test area;

[0008] Building a model predictive controller according to the temperature prediction model and the three-dimensional temperature distribution data, and solving the optimal control problem through the model predictive controller to generate a first temperature control sequence;

[0009] Acquire the operating status data of the refrigeration system and establish a corresponding thermodynamic model, and generate an adaptive refrigeration strategy for the refrigeration system according to the thermodynamic model;

[0010] The first temperature control sequence is controlled and optimized according to the adaptive refrigeration strategy to generate a second temperature control sequence.

[0011] In a second aspect, the present application provides a thermal shock test machine control device, the thermal shock test machine control device comprising:

[0012] An acquisition module, used to acquire the physical structure and material properties of the thermal shock tester and to construct a temperature prediction model of the thermal shock tester;

[0013] A fusion module is used to arrange multiple temperature sensors in the test area and obtain multiple temperature sensor data, and perform adaptive weighted fusion on the multiple temperature sensor data to generate three-dimensional temperature distribution data of the test area;

[0014] A solution module, used for constructing a model prediction controller according to the temperature prediction model and the three-dimensional temperature distribution data, and solving the optimal control problem through the model prediction controller to generate a first temperature control sequence;

[0015] Establishing a module for acquiring the operating status data of the refrigeration system and establishing a corresponding thermodynamic model, and generating an adaptive refrigeration strategy for the refrigeration system according to the thermodynamic model;

[0016] A generating module is used to perform control optimization on the first temperature control sequence according to the adaptive refrigeration strategy to generate a second temperature control sequence.

[0017] The third aspect of the present application provides a computer device, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory so that the computer device executes the above-mentioned hot and cold shock testing machine control method.

[0018] A fourth aspect of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the above-mentioned thermal shock testing machine control method.

[0019] In the technical solution provided by the present application, a target three-dimensional physical model of the test machine is constructed by three-dimensional modeling and material property setting. Based on the finite element model and Fourier equation, a group of heat conduction equations for the transient heat transfer process inside the test machine is established, and a low-dimensional prediction model is formed through dimensionality reduction approximation of the long short-term memory neural network, thereby improving the accuracy of temperature prediction. A multidimensional temperature sensor array is arranged in the test area to obtain multi-point temperature data. Adaptive weighted fusion technology is used to generate high-resolution three-dimensional temperature distribution data to ensure the accuracy and real-time performance of the temperature field distribution. A model predictive controller based on the state space equation is constructed using the temperature prediction model and three-dimensional temperature distribution data. The optimal control problem is solved by the sequential quadratic programming algorithm to generate the future temperature control sequence, thereby achieving precise regulation of temperature control and minimization of energy consumption. Based on the thermodynamic model of the refrigeration system, an adaptive refrigeration strategy is generated. The operating parameters of the refrigeration system, such as evaporation temperature, condensation temperature, compressor frequency and expansion valve opening, are adjusted in real time to improve the energy efficiency ratio and operation stability of the refrigeration system. A heuristic search algorithm is used to optimize the temperature control sequence to ensure the uniformity and accuracy of temperature control. Through refrigeration capacity matching analysis and verification, an optimized temperature control sequence that meets the operation constraints is generated, which improves the control accuracy of the thermal shock testing machine. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0021] Figure 1 A schematic diagram of an embodiment of a control method for a thermal shock testing machine in an embodiment of the present application;

[0022] Figure 2 This is a schematic diagram of an embodiment of a control device for a thermal shock testing machine in an embodiment of the present application. DETAILED DESCRIPTION

[0023] The embodiments of the present application provide a method, device, equipment and storage medium for controlling a hot and cold shock tester. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments described here can be implemented in an order other than that illustrated or described here. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0024] For ease of understanding, the specific process of the embodiment of the present application is described below. Figure 1 , an embodiment of the control method of the thermal shock testing machine in the embodiment of the present application includes:

[0025] Step 101: obtaining the physical structure and material properties of the thermal shock testing machine, and constructing a temperature prediction model of the thermal shock testing machine;

[0026] It is understandable that the execution subject of the present application may be a thermal shock test machine control device, or a terminal or a server, which is not limited here. The present application embodiment is described by taking a server as the execution subject as an example.

[0027] Specifically, the physical structure of the hot and cold shock test machine is modeled in three dimensions to obtain an initial three-dimensional geometric model. According to the material properties of the hot and cold shock test machine, the material parameters of the three-dimensional geometric model are set to obtain the target three-dimensional physical model. The parameters of the materials of different components are defined, including physical properties such as thermal conductivity, density, and specific heat capacity, so that the target three-dimensional physical model can truly simulate the actual heat conduction process. The target three-dimensional physical model is meshed to obtain a finite element model composed of multiple tetrahedral units. Based on the finite element model, a three-dimensional Fourier equation is established to describe the transient heat transfer process inside the hot and cold shock test machine. As the basic equation for heat conduction, the three-dimensional Fourier equation is used to describe the time and space changes of the temperature in the test machine. Combined with specific boundary conditions and initial conditions, a complete set of partial differential equations for heat conduction is obtained. After the heat conduction partial differential equations are spatially discretized, the heat conduction ordinary differential equations are obtained. The spatial discretization process is to discretize the continuous spatial domain and convert it into nodes and units for numerical calculation. The heat conduction ordinary differential equations are time discretized, and the continuous changes in time are converted into discrete time steps to obtain the heat conduction algebraic equations. The heat conduction algebraic equations are solved by numerical methods to obtain the temperature response sequence of the hot and cold shock tester. The temperature response sequence describes the temperature distribution of each point in the tester that changes with time under given conditions. Based on the temperature response sequence, a high-dimensional temperature response function describing the dynamic temperature distribution of the hot and cold shock tester is constructed. This function comprehensively reflects the dynamic characteristics of the temperature change inside the tester. The high-dimensional temperature response function is reduced in dimension. The high-dimensional temperature response function is approximated by the long short-term memory neural network (LSTM) to obtain a low-dimensional prediction model. The low-dimensional prediction model is simulated and verified and error analyzed to ensure the accuracy and reliability of the model. Through simulation verification, the difference between the model prediction value and the actual measurement value is compared to evaluate the performance of the model. According to the error analysis results, the parameters of the low-dimensional prediction model are iteratively optimized, and the model parameters are gradually adjusted to minimize the error, and finally the optimized temperature prediction model is obtained.

[0028] Step 102: Arrange multiple temperature sensors in the test area and obtain multiple temperature sensor data, and perform adaptive weighted fusion on the multiple temperature sensor data to generate three-dimensional temperature distribution data of the test area;

[0029] Specifically, M×N×L temperature sensors are arranged in the test area to form a three-dimensional sensor array. These sensors synchronously collect voltage signals of each temperature sensor through a temperature acquisition circuit, and convert these voltage signals into digital temperature data to obtain M×N×L temperature sampling point data. The M×N×L temperature sampling point data are preprocessed to eliminate noise and outliers to obtain temperature sensor data. Multiple temperature sensor data are spatially interpolated to generate initial temperature field distribution data to ensure the continuity and consistency of temperature data in space. The initial temperature field distribution data is transformed in the frequency domain to obtain the frequency domain characteristics of the temperature field. By analyzing the frequency domain characteristics, the multi-scale components of the temperature field are extracted to obtain the multi-scale decomposition results of the temperature field. Multi-scale analysis helps to capture different frequency components of temperature changes and provide richer temperature field information. The multi-scale decomposition results are adaptively weighted to determine the weight coefficients of different scale components. Based on the weight coefficient, the multi-scale decomposition results are weighted fused to obtain fused temperature field distribution data. The fused temperature field distribution data is spatially reduced to obtain the temperature field feature vector. These eigenvectors represent the main change patterns of the temperature field, reducing the data dimension while retaining the main information of the temperature field. The temperature field eigenvectors are classified, and spatial distribution information is generated based on the classification results. According to the spatial distribution information, the test area is temperature partitioned into P temperature sub-areas. The temperature distribution in each sub-area is relatively uniform, which is conducive to the optimization analysis of the local temperature field. The local temperature field distribution is optimized for each temperature sub-area to ensure that the temperature distribution is more accurate and smooth in each sub-area, and the sub-area temperature distribution data is obtained. Finally, the sub-area temperature distribution data is spliced, and the data of all sub-areas are integrated to generate the three-dimensional temperature distribution data of the entire test area.

[0030] Step 103: construct a model predictive controller according to the temperature prediction model and the three-dimensional temperature distribution data, and solve the optimal control problem through the model predictive controller to generate a first temperature control sequence;

[0031] Specifically, the temperature prediction model and the three-dimensional temperature distribution data are input into the model predictive controller, and the state space equation based on the temperature prediction model is constructed to obtain the first state space model describing the temperature dynamic characteristics of the hot and cold shock test machine. The state space model characterizes the temperature change of the test machine under different operating conditions. According to the first state space model, the first objective function of the model predictive controller is established. The first objective function includes a temperature tracking error term and a control energy consumption term, wherein the temperature tracking error term is used to measure the deviation between the actual temperature and the expected temperature, and the control energy consumption term reflects the energy usage in the control process. By minimizing the first objective function as the optimization goal, the model predictive control problem is constructed to achieve the accuracy of temperature control and the efficiency of energy utilization. The model predictive control problem is solved by using a sequential quadratic programming algorithm. The algorithm can quickly solve the optimal control instruction at each prediction moment and obtain the candidate temperature control sequence for the next N moments. These candidate temperature control sequences contain specific control instructions for the refrigeration system and the heating system, including the refrigeration system power control instruction and the heating system power control instruction. After obtaining these instructions, they are corrected online to adapt to various uncertainties and disturbances in actual operation. Through online correction, the control accuracy is improved to ensure the stability and reliability of the temperature control system in actual operation. After the online correction, a second state space model is constructed to reflect the changes in the system state after the control instructions are corrected. Through the second state space model, the control parameters of the model predictive controller are adaptively adjusted so that the control parameters can dynamically adapt to the actual operation of the system. At the same time, the candidate temperature control sequence is rolled optimized, and the temperature control strategy is always kept in the optimal state by continuously updating the optimization results. Finally, the result of the rolling optimization is the first temperature control sequence, which will be used in the actual temperature control process.

[0032] Step 104: Acquire the operating status data of the refrigeration system and establish a corresponding thermodynamic model, and generate an adaptive refrigeration strategy for the refrigeration system according to the thermodynamic model;

[0033] Specifically, the initial state data of the refrigeration system is obtained, including parameters such as pressure, temperature, and flow rate of the refrigerant. The initial state data is preprocessed to remove noise and outliers to ensure the accuracy and reliability of the data. After preprocessing, the operating state data is obtained. According to the operating state data, a data-driven model of the refrigeration system is constructed. The model captures the operating characteristics of the refrigeration system under different working conditions by analyzing the changing rules of the operating state data. Based on the data-driven model, a thermodynamic model describing the thermodynamic characteristics of the refrigeration system is established. The thermodynamic model integrates the thermodynamic properties of the refrigerant, the heat and mass transfer process of the system, and the operating characteristics of each component, and can accurately simulate the performance of the refrigeration system under different operating conditions. Based on the thermodynamic model, the optimal control problem of the refrigeration system is constructed. The optimal control problem takes the energy efficiency ratio of the refrigeration system as the second objective function, and the constraints include the cooling capacity demand, the compressor frequency range, the expansion valve opening range, etc. By optimizing the control model, the energy efficiency ratio of the refrigeration system can be maximized under the premise of meeting the cooling demand, and the operating efficiency and economy of the system can be improved. In order to solve the optimization control problem, the refrigeration optimization control model is solved, and the optimization algorithm such as the sequential quadratic programming algorithm is used to efficiently find the optimal solution. By solving the refrigeration optimization control model, the optimal operating parameters of the refrigeration system are obtained. These optimal operating parameters include the optimal evaporation temperature, the optimal condensation temperature, the optimal compressor frequency, and the optimal expansion valve opening. The optimal evaporation temperature and the optimal condensation temperature can ensure the best heat transfer effect of the refrigerant in the evaporator and the condenser, while the optimal compressor frequency and the optimal expansion valve opening ensure the efficient operation of the compressor and the expansion valve. After obtaining the optimal operating parameters, the adaptive refrigeration strategy of the refrigeration system is generated in combination with the real-time temperature requirements of the test area. This strategy dynamically adjusts the operating parameters of the refrigeration system by monitoring the temperature changes in the test area in real time, so that the system can quickly respond to changes in temperature requirements and keep the temperature of the test area within the set range. At the same time, the adaptive refrigeration strategy can also automatically adjust various control parameters according to changes in environmental conditions and system status to ensure that the system always operates in the optimal state.

[0034] Step 105: Optimize the control of the first temperature control sequence according to the adaptive refrigeration strategy to generate a second temperature control sequence.

[0035] Specifically, according to the adaptive refrigeration strategy, the control period and optimization amplitude to be optimized in the first temperature control sequence are determined, and the optimization interval and optimization target of the temperature control sequence are obtained. In the optimization interval, the heuristic search algorithm is used to optimize the first temperature control sequence. The heuristic search algorithm can find an approximate optimal solution in a short time by simulating human intuition and experience. In the optimization search process, the search variables are the temperature setting value and duration at each moment, and the third objective function is set as the temperature uniformity and control accuracy in the test area. By optimizing the temperature setting value and the control period, the control accuracy is improved while ensuring the temperature uniformity, and multiple temperature control subsequences are obtained. Multiple temperature control subsequences are merged and spliced ​​to form a spliced ​​temperature control sequence. Through merging and splicing, each optimized subsequence is integrated into a continuous control sequence to meet the needs of actual operation. According to the adaptive refrigeration strategy, the refrigeration capacity matching analysis of the spliced ​​temperature control sequence is performed to obtain refrigeration capacity matching data. By evaluating the matching degree between the spliced ​​sequence and the actual capacity of the refrigeration system, it is ensured that the optimized temperature control sequence will not exceed the capacity range of the refrigeration system in actual operation. According to the refrigeration capacity matching data, the control instructions that exceed the refrigeration capacity range in the spliced ​​temperature control sequence are adjusted and trimmed to generate a temperature control sequence under the refrigeration capacity constraint. Ensure that the final temperature control sequence operates within the capacity of the refrigeration system, thereby avoiding system instability or damage caused by overload operation. Finally, the temperature control sequence under the refrigeration capacity constraint is compared and verified with the operating constraints of the hot and cold shock tester to obtain the verification results. The operating constraints include the physical limitations of the equipment, the safe operating range, etc. Through comparison and verification, it is ensured that the optimized temperature control sequence not only meets the capacity requirements of the refrigeration system, but also meets the various operating constraints of the test machine. According to the verification results, make necessary adjustments to the temperature control sequence to generate the final second temperature control sequence.

[0036] Discretize the first temperature control sequence in the optimization interval. Decompose the continuous temperature control sequence into several temperature control nodes, each of which contains two attributes, namely, the temperature setting value and the duration. Discretization processing helps to simplify the subsequent optimization calculation and provide a clear optimization search space. According to the number of nodes in the discrete temperature control sequence, determine the population size and number of iterations of the heuristic search algorithm, and initialize the parameter settings of the heuristic search algorithm. The determination of the population size and the number of iterations requires a trade-off between computing resources and optimization effects to ensure that the algorithm can converge to a better solution within a reasonable time. Randomly generate an initial population, each individual corresponds to a candidate control subsequence, and the genes of these individuals are composed of the temperature setting value and duration of the temperature control node. For each individual in the initial population, input the candidate control subsequence into the finite element model of the hot and cold shock testing machine, and simulate the three-dimensional temperature distribution change process of the test area under the candidate control subsequence through finite element simulation calculation. Through simulation, the temperature uniformity objective function value and the constraint violation data of the control accuracy corresponding to each candidate control subsequence are obtained. Temperature uniformity reflects the uniformity of temperature distribution in the test area, while control accuracy measures the deviation between the actual temperature control effect and the expected target. After obtaining the objective function value and constraint violation data, the weighted summation method is used to calculate the fitness value of each individual. The higher the fitness value, the better the optimization performance of the candidate control subsequence corresponding to the individual. The weighted summation method combines temperature uniformity and control accuracy by comprehensively considering the two, and effectively combines the two to evaluate the overall pros and cons of each candidate solution. The individuals in the population are selected, crossover and mutated to generate a new generation of population. The selection operation selects individuals with higher fitness values ​​for reproduction according to the principle of survival of the fittest; the crossover operation generates new candidate solutions through gene recombination; the mutation operation introduces new search directions by randomly modifying individual genes to increase the diversity of the population. Through continuous iteration, the algorithm gradually approaches the optimal solution. The iterative process will continue until the preset number of iterations is reached or the convergence condition is met, and finally multiple temperature control subsequences are obtained.

[0037] In the embodiment of the present application, the target three-dimensional physical model of the test machine is constructed by three-dimensional modeling and material property setting. Based on the finite element model and Fourier equation, the heat conduction equation group of the transient heat transfer process inside the test machine is established, and the low-dimensional prediction model is formed by the dimensionality reduction approximation of the long short-term memory neural network, which improves the accuracy of temperature prediction. A multidimensional temperature sensor array is arranged in the test area to obtain multi-point temperature data. Adaptive weighted fusion technology is adopted to generate high-resolution three-dimensional temperature distribution data to ensure the accuracy and real-time performance of the temperature field distribution. A model predictive controller based on the state space equation is constructed using the temperature prediction model and three-dimensional temperature distribution data. Through the sequential quadratic programming algorithm, the optimal control problem is solved and the future temperature control sequence is generated, which realizes the precise adjustment of temperature control and the minimization of energy consumption. Based on the thermodynamic model of the refrigeration system, an adaptive refrigeration strategy is generated. The operating parameters of the refrigeration system, such as evaporation temperature, condensation temperature, compressor frequency and expansion valve opening, are adjusted in real time to improve the energy efficiency ratio and operation stability of the refrigeration system. The temperature control sequence is optimized by using a heuristic search algorithm to ensure the uniformity and accuracy of temperature control. Through refrigeration capacity matching analysis and verification, an optimized temperature control sequence that meets the operation constraints is generated, which improves the control accuracy of the thermal shock testing machine.

[0038] In a specific embodiment, the process of executing step 101 may specifically include the following steps:

[0039] (1) Perform three-dimensional modeling on the physical structure of the thermal shock testing machine to obtain an initial three-dimensional geometric model, and set material parameters for the three-dimensional geometric model according to the material properties of the thermal shock testing machine to obtain a target three-dimensional physical model;

[0040] (2) Meshing the target three-dimensional physical model to obtain a finite element model composed of multiple tetrahedral units;

[0041] (3) Based on the finite element model, a three-dimensional Fourier equation describing the transient heat transfer process inside the thermal shock testing machine was established, and the heat conduction partial differential equation group was obtained by combining the boundary conditions and initial conditions;

[0042] (4) Discretize the heat conduction partial differential equations in space to obtain the heat conduction ordinary differential equations, and discretize the heat conduction ordinary differential equations in time to obtain the heat conduction algebraic equations;

[0043] (5) Solve the heat conduction algebraic equations to obtain the temperature response sequence of the thermal shock test machine. Based on the temperature response sequence, construct a high-dimensional temperature response function that describes the dynamic temperature distribution of the thermal shock test machine.

[0044] (6) Use long short-term memory neural network to reduce the dimension of high-dimensional temperature response function to obtain a low-dimensional prediction model, and perform simulation verification and error analysis on the low-dimensional prediction model to obtain error analysis results;

[0045] (7) Based on the error analysis results, the parameters of the low-dimensional prediction model are iteratively optimized to obtain the temperature prediction model.

[0046] Specifically, the physical structure of the hot and cold shock test machine is three-dimensionally modeled. Using CAD software, such as SolidWorks or AutoCAD, the initial three-dimensional geometric model of the test machine is generated through measurement and design. According to the material properties of the test machine, the material parameters of the three-dimensional geometric model are set. The material properties of each component, including parameters such as thermal conductivity, density and specific heat capacity, will affect the simulation accuracy of the heat conduction process. Through parameter setting, the initial three-dimensional geometric model is converted into a target three-dimensional physical model. After obtaining the target three-dimensional physical model, it is meshed to generate a finite element model composed of multiple tetrahedral units. Meshing is to discretize the continuous three-dimensional space so that it can be solved using numerical methods. Finite element analysis software, such as ANSYS or COMSOL Multiphysics, is usually used for meshing. The density and quality of the mesh directly affect the accuracy and efficiency of the calculation. A denser mesh can provide higher accuracy, but it also increases the amount of calculation. According to the finite element model, a three-dimensional Fourier equation is established to describe the transient heat transfer process inside the hot and cold shock test machine. The Fourier equation describes the diffusion process of heat inside an object, and its general form is:

[0047] ;

[0048] in, Indicates temperature, Indicates time, is the thermal diffusivity, is the thermal conductivity, is the density, is the specific heat capacity. The Laplace operator representing the temperature field. When establishing the Fourier equation, boundary conditions and initial conditions are combined to reflect the actual operating environment of the test machine. For example, the boundary condition can be set as the heat exchange between the outer surface of the test machine and the ambient temperature, and the initial condition is the temperature distribution of the system at startup. The heat conduction partial differential equations are spatially discretized to obtain the heat conduction ordinary differential equations. Spatial discretization is the conversion of continuous space into values ​​at discrete nodes, which is usually achieved through finite difference method, finite volume method or finite element method. The form of the discretized equations is:

[0049] ;

[0050] in, and Represent discrete nodes and The temperature at is the thermal diffusivity. The ordinary differential equations are further discretized in time, and the time domain is also discretized to obtain the heat conduction algebraic equations:

[0051] ;

[0052] in, Indicates time step, the node The temperature, is the time step. By solving the algebraic equations, the temperature response sequence of the hot and cold shock tester is obtained. These sequences describe the temperature distribution of the system at each moment. Based on the temperature response sequence, a high-dimensional temperature response function describing the dynamic temperature distribution of the hot and cold shock tester is constructed. The long short-term memory neural network (LSTM) is used to reduce the dimension of the high-dimensional temperature response function. The LSTM neural network can process time series data, capture the dynamic characteristics of the temperature distribution, and obtain a low-dimensional prediction model by training the network. This low-dimensional model can effectively simplify the calculation while maintaining high-precision prediction of temperature changes. The low-dimensional prediction model is simulated and verified and error analyzed to evaluate the accuracy and reliability of the model. Simulation verification is to compare the prediction model with the actual measurement data. The difference between the model and the actual situation can be found through error analysis. For example, the mean square error (MSE) can be used as an error indicator, which is defined as:

[0053] ;

[0054] in, and Represent the predicted and actual temperature values, respectively. is the total number of samples. By calculating MSE, the error of the model is quantified and the error analysis results are obtained. According to the error analysis results, the parameters of the low-dimensional prediction model are iteratively optimized. The optimization process gradually reduces the prediction error and improves the accuracy of the model by adjusting the model parameters. For example, the gradient descent algorithm can be used to gradually update the model parameters by calculating the gradient of the error:

[0055] ;

[0056] in, For the The model parameters at the iteration, is the learning rate, is the gradient of the error to the parameter. After multiple iterations, the optimized temperature prediction model is finally obtained.

[0057] In a specific embodiment, the process of executing step 102 may specifically include the following steps:

[0058] (1) Arrange M×N×L temperature sensors in the test area to form a three-dimensional sensor array, and synchronously collect the voltage signals of each temperature sensor through the temperature acquisition circuit, and convert the voltage signals into digital temperature data to obtain M×N×L temperature sampling point data;

[0059] (2) Preprocessing the data of M×N×L temperature sampling points to obtain multiple temperature sensor data, and performing spatial interpolation on the multiple temperature sensor data to obtain initial temperature field distribution data;

[0060] (3) Perform frequency domain transformation on the initial temperature field distribution data to obtain the frequency domain characteristics of the temperature field, and extract the multi-scale components of the temperature field based on the frequency domain characteristics to obtain the multi-scale decomposition results of the temperature field;

[0061] (4) Adaptively weight the multi-scale decomposition results to obtain weight coefficients of components of different scales, and perform weighted fusion on the multi-scale decomposition results according to the weight coefficients to obtain fused temperature field distribution data;

[0062] (5) Perform spatial dimension reduction on the fused temperature field distribution data to obtain the temperature field feature vector, and classify the temperature field feature vector to obtain spatial distribution information;

[0063] (6) Based on the spatial distribution information, the test area is partitioned into temperature zones to obtain P temperature sub-regions, and the local temperature field distribution of each temperature sub-region is optimized to obtain the sub-region temperature distribution data;

[0064] (7) The temperature distribution data of the sub-regions are spliced ​​to obtain the three-dimensional temperature distribution data of the test area.

[0065] Specifically, temperature sensors are arranged in the test area to form a three-dimensional sensor array. Suppose there is a laboratory with a test area of ​​2×2×2 meters. A temperature sensor is arranged every 0.5 meters. There will be 5 sensors in each dimension, for a total of 5×5×5=125 temperature sensors. These sensors synchronously collect their respective voltage signals through the temperature acquisition circuit. The voltage signal is the response of the sensor to temperature changes. Usually, temperature sensors such as thermocouples or RTD sensors output a voltage signal that is proportional to the temperature. Use an analog-to-digital converter to convert these analog voltage signals into digital temperature data. Assuming a 12-bit analog-to-digital converter is used, the voltage signal of each sensor will be converted into a digital value between 0 and 4095. After obtaining the digital temperature data, The data of the temperature sampling points are preprocessed. The preprocessing step includes removing noise and outliers. Assuming that the data of a certain sensor deviates significantly from the data of other sensors, use median filtering or mean filtering to smooth the data. Through preprocessing, more accurate and consistent data of multiple temperature sensors are obtained. Spatial interpolation is performed on the data of multiple temperature sensors to obtain the initial temperature field distribution data. There are many methods of spatial interpolation, such as inverse distance weighted (IDW) or Kriging interpolation. Assuming the IDW interpolation method is used, its basic formula is:

[0066] ;

[0067] in, is the interpolation point The temperature at It is a known point The temperature at is the interpolation point To a known point The distance is the weight factor. Through interpolation calculation, the initial temperature field distribution data of the entire test area is obtained. The initial temperature field distribution data is transformed in the frequency domain to obtain the frequency domain characteristics of the temperature field. The frequency domain transformation is usually implemented using the fast Fourier transform, and its formula is:

[0068] ;

[0069] in, is the frequency domain signal, is a time domain signal, is the signal length, is the frequency index. Through fast Fourier transform, the initial temperature field distribution data is converted from the time domain (or space domain) to the frequency domain to reveal the frequency components of temperature change. According to the frequency domain characteristics, the multi-scale components of the temperature field are extracted to obtain the multi-scale decomposition results of the temperature field. Wavelet transform is a commonly used multi-scale analysis method, and its basic formula is:

[0070] ;

[0071] in, are the wavelet transform coefficients, It's a signal. is the mother wavelet function, is the scale parameter, is the translation parameter. Through multi-scale decomposition, the different frequency components of the temperature field are separated. The multi-scale decomposition results are adaptively weighted to obtain the weight coefficients of the different scale components. The weight coefficients can be set according to the importance of each scale component in the temperature field. For example, an adaptive method is used to optimize these coefficients so that the weighted fusion results are most consistent with the actual situation. The weighting formula is:

[0072] ;

[0073] in, is the weighted fusion temperature, It is The weight of the scale component, It is The temperature of the scale component. Perform spatial dimensionality reduction on the fused temperature field distribution data to obtain the temperature field feature vector. Principal component analysis is a commonly used dimensionality reduction method that maps high-dimensional data to low-dimensional space through linear transformation while retaining the data variance as much as possible. Classify the temperature field feature vector to obtain spatial distribution information. Cluster analysis is a commonly used classification method, such as K-means clustering, which divides the data into The data similarity within each cluster is the largest, and the data similarity between different clusters is the smallest. The basic formula of K-means clustering is:

[0074] ;

[0075] in, is the objective function, is the number of clusters, It is Clusters, It is data points, It is According to the spatial distribution information, the test area is partitioned into temperature zones and the temperature sub-regions. Through this partitioning, the entire test area is divided into multiple relatively independent sub-regions, and the temperature distribution in each sub-region is more uniform and stable. The local temperature field distribution of each temperature sub-region is optimized to obtain the sub-region temperature distribution data. Various methods can be used for local optimization, such as genetic algorithms, particle swarm algorithms, etc. Through the optimization algorithm, the optimal temperature control strategy in each sub-region is found to ensure the uniformity and stability of the temperature distribution. The sub-region temperature distribution data are spliced ​​to obtain the three-dimensional temperature distribution data of the test area. The splicing process needs to ensure that the temperature data between different sub-regions transitions smoothly at the boundary to avoid temperature mutations or discontinuities.

[0076] In a specific embodiment, the process of executing step 103 may specifically include the following steps:

[0077] (1) The temperature prediction model and three-dimensional temperature distribution data are input into the model predictive controller, and a state space equation based on the temperature prediction model is constructed to obtain the first state space model describing the temperature dynamic characteristics of the hot and cold shock testing machine;

[0078] (2) According to the first state space model, a first objective function of the model predictive controller is established. The first objective function includes a temperature tracking error term and a control energy consumption term. The model predictive control problem is constructed by minimizing the first objective function as the optimization goal;

[0079] (3) The model predictive control problem is solved using a sequential quadratic programming algorithm to obtain candidate temperature control sequences for the next N moments, and based on the candidate temperature control sequences, corresponding cooling system power control instructions and heating system power control instructions are generated;

[0080] (4) performing online correction on the cooling system power control command and the heating system power control command to obtain a second state space model;

[0081] (5) The control parameters of the model predictive controller are adaptively adjusted through the second state space model, and the candidate temperature control sequence is rolled optimized to obtain the first temperature control sequence.

[0082] Specifically, the temperature prediction model and the three-dimensional temperature distribution data are integrated. The temperature prediction model is based on previous data-driven and physical modeling, and can predict the temperature changes of the system under different control instructions. The three-dimensional temperature distribution data provides the temperature state of each position inside the test machine at the current moment. Based on these data, the state space equation is constructed. The state space model is a method to describe a dynamic system that can link the input, output, and state of the system together. The general form of the state space equation is:

[0083] ;

[0084] in, Indicates that the system is at time The state vector of represents the control input, Indicates the output, is the system matrix, and Represent process noise and measurement noise respectively. According to the first state space model, the first objective function of the model predictive controller is established. The objective function includes a temperature tracking error term and a control energy consumption term. The temperature tracking error term is used to measure the deviation between the actual temperature and the set temperature, and the control energy consumption term reflects the energy usage in the control process. The objective function can be expressed as:

[0085] ;

[0086] in, is the objective function, is the reference temperature, and is the weight matrix, Used to trade off temperature tracking error, Used to balance control energy consumption. With minimizing the first objective function as the optimization goal, the model predictive control problem is constructed. By optimizing the temperature control sequence, the objective function is minimized, thereby achieving accurate control of temperature and effective use of energy. The sequential quadratic programming algorithm (SQP) is used to solve the model predictive control problem. SQP is an effective nonlinear optimization algorithm that can handle constraints and obtain the optimal solution. Through the SQP solution, the candidate temperature control sequence for the next N moments is obtained. This control sequence is the control strategy of the system in the future, which includes the temperature set value and the corresponding control instructions at each moment. According to the candidate temperature control sequence, the corresponding refrigeration system power control instructions and heating system power control instructions are generated. These instructions are used to adjust the power output of the refrigeration and heating systems to achieve accurate temperature control. The refrigeration system power control instructions and the heating system power control instructions are corrected online. The online correction is based on real-time temperature feedback and system status to ensure that the control instructions can adapt to the current system status. The corrected control instructions are used to update the second state space model to make it more accurately reflect the dynamic characteristics of the system. Through the second state space model, the control parameters of the model predictive controller are adaptively adjusted. Adaptive adjustment continuously updates the control parameters by analyzing the system's operating data in real time, so that it can adapt to changes in the system. At the same time, the candidate temperature control sequence is optimized in a rolling manner. Rolling optimization is an online optimization method that can update the control strategy at each moment based on the latest system status and prediction information to ensure that the system always operates in the optimal state. Finally, the first temperature control sequence is obtained. For example, suppose there is a hot and cold shock test machine that needs to quickly drop the temperature from room temperature to -40°C during the test, keep it for a period of time, and then quickly rise to 85°C. Input the temperature prediction model and the current three-dimensional temperature distribution data into the model predictive controller to construct the state space equation. Set the objective function, in which the temperature tracking error term measures the deviation between the actual temperature and the target temperature, and the control energy consumption term reflects the energy usage of the refrigeration and heating system. Solve the model predictive control problem through the SQP algorithm to obtain the candidate temperature control sequence for a period of time in the future. Assume that the obtained sequence is as follows:

[0087] ;

[0088] Each tuple represents a temperature setting value and duration. Based on this control sequence, the corresponding power control instructions for the cooling system and the heating system are generated. For example, in the first time period, the power of the cooling system is set to 50% for 5 minutes; in the second time period, the power of the cooling system is set to 100% for 10 minutes. During the control process, the power control instructions for the cooling and heating systems are corrected online by real-time monitoring of the system temperature. For example, if the temperature drops too fast, the power output of the cooling system can be appropriately reduced; if the temperature rises too slowly, the power output of the heating system can be increased. Through these corrections, it is ensured that the system temperature can accurately track the set temperature. At the same time, the control parameters are adaptively adjusted through the second state space model. For example, according to the real-time temperature data, the heat transfer coefficient and time constant in the model are adjusted to make the model more accurately reflect the dynamic characteristics of the system. Finally, through rolling optimization, the candidate temperature control sequence is updated to ensure that the system always runs in the optimal state.

[0089] In a specific embodiment, the process of executing step 104 may specifically include the following steps:

[0090] (1) Obtaining the initial state data of the refrigeration system and preprocessing the initial state data to obtain the operating state data;

[0091] (2) Based on the operating status data, a data-driven model of the refrigeration system is constructed, and based on the data-driven model, a thermodynamic model describing the thermodynamic characteristics of the refrigeration system is established;

[0092] (3) Based on the thermodynamic model, the optimal control problem of the refrigeration system is constructed, where the second objective function is the energy efficiency ratio of the refrigeration system, and the constraints include the cooling capacity demand, the compressor frequency range, and the expansion valve opening range, and the refrigeration optimization control model is obtained;

[0093] (4) Solve the refrigeration optimization control model to obtain the optimal operating parameters of the refrigeration system, which include the optimal evaporation temperature, the optimal condensation temperature, the optimal compressor frequency and the optimal expansion valve opening;

[0094] (5) Generate an adaptive refrigeration strategy for the refrigeration system based on the optimal operating parameters and the real-time temperature requirements of the test area.

[0095] Specifically, the initial state data is obtained from various sensors and measuring devices of the refrigeration system. These data include parameters such as pressure, temperature, and flow of the refrigerant, as well as control variables such as the frequency of the compressor and the opening of the expansion valve. It is assumed that the temperature and pressure data of the refrigerant at the inlet and outlet of the evaporator are collected through the sensors installed in the refrigeration system, and the working state data of the compressor and the expansion valve are also collected. After obtaining the initial state data, these data are preprocessed. The preprocessing step includes removing noise and outliers to ensure the accuracy and consistency of the data. For example, median filtering or mean filtering is used to smooth the data and eliminate occasional noise. The data is standardized to convert data of different dimensions to the same scale for subsequent model construction and optimization. After obtaining the preprocessed operating state data, a data-driven model of the refrigeration system is constructed based on these data. The data-driven model captures the dynamic characteristics of the system by analyzing the relationship between historical operating data and current state data. It is assumed that a regression model in machine learning, such as a multivariate linear regression or a neural network model, is used to construct a data-driven model. The model can predict the response of the refrigeration system under different control inputs. According to the data-driven model, a thermodynamic model describing the thermodynamic characteristics of the refrigeration system is established. Thermodynamic models reveal the energy conversion and heat transfer process of the system by describing the thermodynamic process of the refrigerant in the system. Thermodynamic models are usually based on conservation of mass, conservation of energy, and equation of state. For example, the following formula can be used to describe the thermodynamic process of the refrigerant in the evaporator:

[0096] ;

[0097] in, is the cooling capacity of the evaporator, is the mass flow rate of refrigerant, and are the specific enthalpy of the refrigerant at the outlet and inlet of the evaporator, respectively. In this way, the thermodynamic model of each part of the refrigeration system is established. Based on the thermodynamic model, the optimal control problem of the refrigeration system is constructed. The second objective function of this problem is the energy efficiency ratio (COP) of the refrigeration system, which is defined as the ratio of cooling capacity to power consumption:

[0098] ;

[0099] in, is the cooling capacity of the evaporator, is the power consumption of the compressor. Constraints include cooling capacity demand, compressor frequency range, and expansion valve opening range. For example, the cooling capacity demand can be set to meet a certain temperature range, the compressor frequency can be limited to between 30 Hz and 60 Hz, and the expansion valve opening can be limited to between 10% and 90%. Through these constraints, a refrigeration optimization control model is obtained. The refrigeration optimization control model is solved to obtain the optimal operating parameters of the refrigeration system. Using an optimization algorithm, such as a genetic algorithm or a particle swarm optimization algorithm, the operating parameters that maximize the energy efficiency ratio can be found under the premise of satisfying all constraints. Assume that by solving the optimization model, the optimal operating parameters obtained include: the optimal evaporation temperature is -10°C, the optimal condensing temperature is 40°C, the optimal compressor frequency is 45 Hz, and the optimal expansion valve opening is 50%. Based on the optimal operating parameters and the real-time temperature demand of the test area, an adaptive refrigeration strategy for the refrigeration system is generated. The adaptive refrigeration strategy dynamically adjusts the operating parameters of the refrigeration system by real-time monitoring of the temperature changes in the test area. For example, when the temperature in the test area is higher than the set value, the adaptive strategy can increase the frequency of the compressor or adjust the opening of the expansion valve to improve the cooling capacity; when the temperature is lower than the set value, the frequency of the compressor can be reduced or the opening of the expansion valve can be reduced to reduce the cooling capacity and avoid overcooling.

[0100] In a specific embodiment, the process of executing step 105 may specifically include the following steps:

[0101] (1) According to the adaptive cooling strategy, the control period and optimization range to be optimized in the first temperature control sequence are determined, and the optimization interval and optimization target of the temperature control sequence are obtained;

[0102] (2) In the optimization interval, a heuristic search algorithm is used to optimize the first temperature control sequence. The search variables are the temperature setting value and duration at each moment. The third objective function is the temperature uniformity and control accuracy in the test area. Multiple temperature control subsequences are obtained.

[0103] (3) Merge and splice multiple temperature control subsequences to obtain a spliced ​​temperature control sequence, and perform cooling capacity matching analysis on the spliced ​​temperature control sequence according to the adaptive cooling strategy to obtain cooling capacity matching data;

[0104] (4) According to the refrigeration capacity matching data, the control instructions in the spliced ​​temperature control sequence that exceed the refrigeration capacity range are adjusted and cut to obtain the temperature control sequence under the refrigeration capacity constraint;

[0105] (5) Compare and verify the temperature control sequence under the refrigeration capacity constraint with the operation constraint conditions of the thermal shock testing machine to obtain a verification result, and generate a corresponding second temperature control sequence according to the verification result.

[0106] Specifically, the current temperature control sequence is analyzed to find out the parts that need to be improved. The adaptive refrigeration strategy identifies the time periods when the temperature control effect is not ideal by real-time monitoring and analyzing the temperature changes of the system. These time periods are the control time periods to be optimized. The optimization range is determined by the temperature deviation and control accuracy of the system during these time periods. For example, if the temperature fluctuates greatly or deviates greatly from the set temperature during a certain time period, then this time period requires a larger optimization range. A heuristic search algorithm is used to optimize the first temperature control sequence. The heuristic search algorithm is a search method that simulates the human thinking process. It gradually approaches the optimal solution through continuous trial and adjustment. Genetic algorithm is used as a heuristic search algorithm. The genetic algorithm optimizes the temperature control sequence by simulating the process of natural selection and genetic variation. The first temperature control sequence in the optimization interval is discretized to obtain a discrete temperature control sequence consisting of several temperature control nodes. Each temperature control node includes two attributes: temperature set value and duration. For example, a temperature control node can be represented as ,in is the temperature setting value, is the duration of the temperature. According to the number of nodes in the discrete temperature control sequence, the population size and number of iterations of the genetic algorithm are determined, and the parameter settings of the genetic algorithm are initialized. Assume that the population size is set to 50 and the number of iterations is 1000. In the initial population generation process of the genetic algorithm, each individual corresponds to a candidate control subsequence. The gene of the individual consists of the temperature setting value and duration of the temperature control node. After the initial population is randomly generated, each individual in the population is evaluated. Each candidate control subsequence is input into the finite element model of the hot and cold shock testing machine to simulate the three-dimensional temperature distribution change process of the test area under the sequence. Through simulation, the objective function value reflecting the temperature uniformity and the constraint violation data reflecting the control accuracy are obtained. The objective function can be defined as:

[0107] ;

[0108] in, is the objective function value, and is the weight coefficient, is the temperature at the kth moment, is the reference temperature, To control energy consumption. According to the objective function value and constraint violation data, the weighted summation method is used to calculate the fitness value of each individual. The higher the fitness value, the better the optimization performance of the candidate control subsequence corresponding to the individual. The individuals in the population are selected, crossed and mutated to generate a new generation of population. The selection operation selects individuals with higher fitness values ​​for reproduction according to the principle of survival of the fittest; the crossover operation generates new candidate solutions through gene recombination; the mutation operation introduces new search directions by randomly modifying individual genes to increase the diversity of the population. Through continuous iteration, the algorithm gradually approaches the optimal solution. The iterative process will continue until the preset number of iterations is reached or the convergence condition is met, and multiple temperature control subsequences are obtained. These temperature control subsequences are merged and spliced ​​to obtain a spliced ​​temperature control sequence. According to the adaptive refrigeration strategy, the spliced ​​temperature control sequence is analyzed for refrigeration capacity matching to obtain refrigeration capacity matching data. The matching analysis ensures that the optimized temperature control sequence does not exceed the capacity of the refrigeration system in actual operation by evaluating the matching degree between the spliced ​​sequence and the actual capacity of the refrigeration system. For example, if the maximum cooling capacity of the refrigeration system is 5000W, and the cooling load demand in a certain time period exceeds this value, the control instructions for this time period need to be adjusted. According to the refrigeration capacity matching data, the control instructions that exceed the refrigeration capacity range in the spliced ​​temperature control sequence are adjusted and cut to obtain the temperature control sequence under the refrigeration capacity constraint. By reducing the excessively high temperature setting value or extending the duration, ensure that the control sequence operates within the refrigeration capacity range. Compare and verify the temperature control sequence under the refrigeration capacity constraint with the operating constraints of the thermal shock tester. The operating constraints include the physical limitations of the equipment, the safe operating range, etc. Through comparison and verification, ensure that the optimized temperature control sequence not only meets the capacity requirements of the refrigeration system, but also meets the various operating constraints of the tester. For example, the temperature change rate in the temperature control sequence cannot exceed the allowable range of the equipment, and the temperature setting value cannot exceed the safe operating temperature. Finally, the corresponding second temperature control sequence is generated according to the verification results. This sequence is based on the results of the optimization algorithm, refrigeration capacity matching analysis and operation constraint verification, and can provide accurate and stable temperature control in actual operation.

[0109] In a specific embodiment, the execution step is to optimize the first temperature control sequence within the optimization interval using a heuristic search algorithm, the search variables are the temperature setting value and duration at each moment, the third objective function is the temperature uniformity and control accuracy within the test area, and the process of obtaining multiple temperature control subsequences can specifically include the following steps:

[0110] (1) Discretize the first temperature control sequence within the optimization interval to obtain a discrete temperature control sequence consisting of a number of temperature control nodes, each of which includes two attributes: a temperature setting value and a duration;

[0111] (2) According to the number of nodes in the discrete temperature control sequence, determine the population size and number of iterations of the heuristic search algorithm, and initialize the parameter settings of the heuristic search algorithm;

[0112] (3) Randomly generate an initial population, where each individual corresponds to a candidate control subsequence, and the individual's gene consists of the temperature setting value and duration of the temperature control node;

[0113] (4) For each individual in the initial population, the candidate control subsequence is input into the finite element model of the hot and cold shock testing machine to simulate the three-dimensional temperature distribution change process of the test area under the candidate control subsequence, and the objective function value reflecting the temperature uniformity and the constraint violation data reflecting the control accuracy are obtained;

[0114] (5) Based on the objective function value and constraint violation data, the fitness value of each individual is calculated using the weighted summation method. The higher the fitness value, the better the optimization performance of the candidate control subsequence corresponding to the individual.

[0115] (6) Perform selection, crossover and mutation operations on individuals in the population to generate a new generation of population until the preset number of iterations is reached or the convergence condition is met, and multiple temperature control subsequences are obtained.

[0116] Specifically, the continuous temperature control sequence is discretized in time. Assume that the optimization interval is from time arrive A period of time is divided into several time periods, and the length of each time period is , get discrete time points For each discrete time point, the corresponding temperature setting value and duration are recorded to form a temperature control node. For example, a temperature control node can be represented as ,in Indicates the temperature setting value. Represents the duration. After obtaining the discrete temperature control sequence, the population size and number of iterations of the heuristic search algorithm are determined according to the number of these temperature control nodes. Heuristic search algorithms, such as genetic algorithms, are optimized by simulating the process of natural selection and genetic variation. The population size determines the number of individuals in each generation, while the number of iterations determines the number of cycles the algorithm runs. Assume that there are control nodes, select a population size of 50 and an iteration number of 1000. These parameters need to be adjusted according to the complexity of the problem and the computing resources. After determining the population size and the number of iterations, initialize the parameter settings of the heuristic search algorithm. The parameters of the genetic algorithm include the crossover rate, mutation rate, etc. The crossover rate indicates the probability of gene exchange between two individuals, and the mutation rate indicates the probability of random changes in individual genes. Assuming the crossover rate is 0.8 and the mutation rate is 0.1, these parameters will affect the convergence speed of the algorithm and the quality of the results. Randomly generate the initial population, and each individual corresponds to a candidate control subsequence. The genes of an individual consist of the temperature setting value and duration of the temperature control node. Assuming the initial population size is 50 and the gene length of each individual is , randomly generate 50 candidate control subsequences, each subsequence contains temperature control nodes. For example, an individual can be represented as For each individual in the initial population, the candidate control subsequence is input into the finite element model of the thermal shock test machine to simulate the three-dimensional temperature distribution change process of the test area under the candidate control subsequence. The finite element model simulates the change of temperature over time and space by solving the heat conduction equation. For example, the heat conduction equation can be expressed as:

[0117] ;

[0118] in, Indicates temperature, Indicates time, is the thermal diffusivity, is the thermal conductivity, is the density, is the specific heat capacity. By numerically solving this equation, the temperature distribution change process under different candidate control subsequences is obtained. During the simulation process, the objective function value reflecting the temperature uniformity and the constraint violation data reflecting the control accuracy are obtained. The objective function can be defined as:

[0119] ;

[0120] in, is the objective function value, and is the weight coefficient, For the The temperature at that moment, is the reference temperature, To control energy consumption. Constraint violation data reflects whether the temperature control exceeds the set safety range. According to the objective function value and constraint violation data, the weighted summation method is used to calculate the fitness value of each individual. The higher the fitness value, the better the optimization performance of the candidate control subsequence corresponding to the individual. The fitness value can be defined as:

[0121] Fitness ;

[0122] Among them, Fitness Indicates The fitness value of each individual, For the The objective function value of each individual, is a very small number to prevent the denominator from being zero. The individuals in the population are selected, crossovered, and mutated to generate a new generation of population. The selection operation selects individuals with higher fitness values ​​for reproduction according to the principle of survival of the fittest; the crossover operation generates new candidate solutions through gene recombination; the mutation operation introduces new search directions by randomly modifying individual genes to increase the diversity of the population. For example, in the crossover operation, two individuals are randomly selected and part of their genes are exchanged to generate two new individuals. Through continuous iteration, the algorithm gradually approaches the optimal solution. The iterative process will continue until the preset number of iterations is reached or the convergence condition is met, and finally multiple temperature control subsequences are obtained. These subsequences can provide more accurate and stable temperature control through optimized search.

[0123] The above describes the control method of the hot and cold shock tester in the embodiment of the present application. The following describes the control device of the hot and cold shock tester in the embodiment of the present application. Figure 2 In the embodiment of the present application, one embodiment of the control device of the thermal shock testing machine includes:

[0124] An acquisition module 201 is used to acquire the physical structure and material properties of the thermal shock tester and to construct a temperature prediction model of the thermal shock tester;

[0125] A fusion module 202 is used to arrange multiple temperature sensors in the test area and obtain multiple temperature sensor data, and perform adaptive weighted fusion on the multiple temperature sensor data to generate three-dimensional temperature distribution data of the test area;

[0126] A solution module 203 is used to construct a model prediction controller according to the temperature prediction model and the three-dimensional temperature distribution data, and solve the optimal control problem through the model prediction controller to generate a first temperature control sequence;

[0127] Establishing module 204, for acquiring the operating status data of the refrigeration system and establishing a corresponding thermodynamic model, and generating an adaptive refrigeration strategy for the refrigeration system according to the thermodynamic model;

[0128] The generating module 205 is used to perform control optimization on the first temperature control sequence according to the adaptive refrigeration strategy to generate a second temperature control sequence.

[0129] Through the cooperation of the above components, the target three-dimensional physical model of the test machine was constructed through three-dimensional modeling and material property setting. Based on the finite element model and Fourier equation, the heat conduction equation group of the transient heat transfer process inside the test machine was established, and the low-dimensional prediction model was formed through dimensionality reduction approximation of the long short-term memory neural network, which improved the accuracy of temperature prediction. A multi-dimensional temperature sensor array was arranged in the test area to obtain multi-point temperature data. Adaptive weighted fusion technology was used to generate high-resolution three-dimensional temperature distribution data to ensure the accuracy and real-time distribution of the temperature field. A model predictive controller based on the state space equation was constructed using the temperature prediction model and three-dimensional temperature distribution data. The optimal control problem was solved through the sequential quadratic programming algorithm to generate the future temperature control sequence, realizing the precise adjustment of temperature control and minimization of energy consumption. Based on the thermodynamic model of the refrigeration system, an adaptive refrigeration strategy was generated. The operating parameters of the refrigeration system, such as evaporation temperature, condensation temperature, compressor frequency and expansion valve opening, were adjusted in real time to improve the energy efficiency ratio and operation stability of the refrigeration system. The temperature control sequence was optimized using a heuristic search algorithm to ensure the uniformity and accuracy of temperature control. Through refrigeration capacity matching analysis and verification, an optimized temperature control sequence that meets the operation constraints is generated, which improves the control accuracy of the thermal shock testing machine.

[0130] The present application also provides a computer device, which includes a memory and a processor. The memory stores computer-readable instructions. When the computer-readable instructions are executed by the processor, the processor executes the steps of the hot and cold shock testing machine control method in the above-mentioned embodiments.

[0131] The present application also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions are executed on a computer, the computer executes the steps of the hot and cold shock testing machine control method.

[0132] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0133] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store program codes.

[0134] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A control method for a thermal shock testing machine, characterized in that: The method comprises: The physical structure and material properties of the hot and cold shock testing machine are obtained, and a temperature prediction model of the hot and cold shock testing machine is constructed; specifically, the method comprises: performing three-dimensional modeling on the physical structure of the hot and cold shock testing machine to obtain an initial three-dimensional geometric model, and setting material parameters of the three-dimensional geometric model according to the material properties of the hot and cold shock testing machine to obtain a target three-dimensional physical model; meshing the target three-dimensional physical model to obtain a finite element model composed of a plurality of tetrahedral units; according to the finite element model, establishing a three-dimensional Fourier equation describing the transient heat transfer process inside the hot and cold shock testing machine, and combining boundary conditions and initial conditions to obtain a group of partial differential equations for heat conduction; and meshing the target three-dimensional physical model to obtain a finite element model composed of a plurality of tetrahedral units. The differential equations are spatially discretized to obtain a heat conduction ordinary differential equation group, and the heat conduction ordinary differential equation group is time discretized to obtain a heat conduction algebraic equation group; the heat conduction algebraic equation group is solved to obtain a temperature response sequence of the hot and cold shock tester, and based on the temperature response sequence, a high-dimensional temperature response function describing the dynamic temperature distribution of the hot and cold shock tester is constructed; the high-dimensional temperature response function is approximated by dimensionality reduction using a long short-term memory neural network to obtain a low-dimensional prediction model, and the low-dimensional prediction model is simulated and verified and error analyzed to obtain an error analysis result; according to the error analysis result, the low-dimensional prediction model is iteratively optimized in parameters to obtain a temperature prediction model; Arranging a plurality of temperature sensors in the test area and acquiring a plurality of temperature sensor data, and adaptively weighting and fusing the plurality of temperature sensor data to generate three-dimensional temperature distribution data of the test area; Building a model predictive controller according to the temperature prediction model and the three-dimensional temperature distribution data, and solving the optimal control problem through the model predictive controller to generate a first temperature control sequence; Acquire the operating status data of the refrigeration system and establish a corresponding thermodynamic model, and generate an adaptive refrigeration strategy for the refrigeration system according to the thermodynamic model; The first temperature control sequence is controlled and optimized according to the adaptive refrigeration strategy to generate a second temperature control sequence.

2. The thermal shock testing machine control method according to claim 1, characterized in that: The method of arranging a plurality of temperature sensors in the test area and acquiring a plurality of temperature sensor data, and adaptively weighting and fusing the plurality of temperature sensor data to generate three-dimensional temperature distribution data of the test area includes: Arrange M×N×L temperature sensors in the test area to form a three-dimensional sensor array, and synchronously collect voltage signals of each temperature sensor through a temperature acquisition circuit, and convert the voltage signals into digital temperature data to obtain M×N×L temperature sampling point data; Preprocessing the M×N×L temperature sampling point data to obtain a plurality of temperature sensor data, and performing spatial interpolation on the plurality of temperature sensor data to obtain initial temperature field distribution data; Performing frequency domain transformation on the initial temperature field distribution data to obtain frequency domain characteristics of the temperature field, and extracting multi-scale components of the temperature field according to the frequency domain characteristics to obtain a multi-scale decomposition result of the temperature field; Adaptively weighting the multi-scale decomposition results to obtain weight coefficients of components of different scales, and weighted fusion of the multi-scale decomposition results according to the weight coefficients to obtain fused temperature field distribution data; Performing spatial dimension reduction on the fused temperature field distribution data to obtain temperature field feature vectors, and classifying the temperature field feature vectors to obtain spatial distribution information; According to the spatial distribution information, the test area is temperature partitioned to obtain P temperature sub-areas, and local temperature field distribution optimization is performed on each temperature sub-area to obtain sub-area temperature distribution data; The sub-region temperature distribution data are spliced ​​to obtain three-dimensional temperature distribution data of the test region.

3. The thermal shock testing machine control method according to claim 2, characterized in that: The step of constructing a model predictive controller according to the temperature prediction model and the three-dimensional temperature distribution data, and solving the optimal control problem by means of the model predictive controller to generate a first temperature control sequence comprises: Inputting the temperature prediction model and the three-dimensional temperature distribution data into a model prediction controller, constructing a state space equation based on the temperature prediction model, and obtaining a first state space model describing the temperature dynamic characteristics of the thermal shock testing machine; According to the first state-space model, a first objective function of the model predictive controller is established, wherein the first objective function includes a temperature tracking error term and a control energy consumption term, and a model predictive control problem is constructed by minimizing the first objective function as an optimization goal; A sequential quadratic programming algorithm is used to solve the model predictive control problem to obtain a candidate temperature control sequence for N future moments, and corresponding cooling system power control instructions and heating system power control instructions are generated according to the candidate temperature control sequence; Performing online correction on the refrigeration system power control command and the heating system power control command to obtain a second state space model; The control parameters of the model predictive controller are adaptively adjusted through the second state space model, and the candidate temperature control sequence is rolling optimized to obtain a first temperature control sequence.

4. The method for controlling a thermal shock testing machine according to claim 1, characterized in that: The step of acquiring the operating status data of the refrigeration system and establishing a corresponding thermodynamic model, and generating an adaptive refrigeration strategy for the refrigeration system according to the thermodynamic model, comprises: Acquiring initial state data of the refrigeration system, and preprocessing the initial state data to obtain operating state data; constructing a data-driven model of the refrigeration system according to the operating status data, and establishing a thermodynamic model describing the thermodynamic characteristics of the refrigeration system according to the data-driven model; Based on the thermodynamic model, the optimal control problem of the refrigeration system is constructed, wherein the second objective function is the energy efficiency ratio of the refrigeration system, and the constraints include the refrigeration capacity demand, the compressor frequency range, and the expansion valve opening range, thereby obtaining a refrigeration optimization control model; Solving the refrigeration optimization control model to obtain optimal operating parameters of the refrigeration system, wherein the optimal operating parameters include an optimal evaporation temperature, an optimal condensation temperature, an optimal compressor frequency, and an optimal expansion valve opening; An adaptive refrigeration strategy for the refrigeration system is generated according to the optimal operating parameters and in combination with the real-time temperature demand of the test area.

5. The method for controlling a thermal shock testing machine according to claim 1, characterized in that: The step of controlling and optimizing the first temperature control sequence according to the adaptive refrigeration strategy to generate a second temperature control sequence includes: According to the adaptive refrigeration strategy, determining the control period and optimization range to be optimized in the first temperature control sequence, and obtaining the optimization interval and optimization target of the temperature control sequence; In the optimization interval, the first temperature control sequence is optimized and searched by using a heuristic search algorithm, the search variables are the temperature setting value and the duration at each moment, and the third objective function is the temperature uniformity and control accuracy in the test area, to obtain a plurality of temperature control subsequences; Merging and splicing the multiple temperature control subsequences to obtain a spliced ​​temperature control sequence, and performing a refrigeration capacity matching analysis on the spliced ​​temperature control sequence according to the adaptive refrigeration strategy to obtain refrigeration capacity matching degree data; According to the refrigeration capacity matching data, control instructions in the spliced ​​temperature control sequence that exceed the refrigeration capacity range are adjusted and cut to obtain a temperature control sequence under refrigeration capacity constraints; The temperature control sequence under the refrigeration capacity constraint is compared and verified with the operation constraint condition of the thermal shock testing machine to obtain a verification result, and a corresponding second temperature control sequence is generated according to the verification result.

6. The thermal shock testing machine control method according to claim 5, characterized in that: In the optimization interval, the first temperature control sequence is optimized and searched by using a heuristic search algorithm, the search variables are the temperature setting value and duration at each moment, the third objective function is the temperature uniformity and control accuracy in the test area, and multiple temperature control subsequences are obtained, including: Discretizing the first temperature control sequence within the optimization interval to obtain a discrete temperature control sequence consisting of a plurality of temperature control nodes, each temperature control node including two attributes: a temperature setting value and a duration; Determining the population size and number of iterations of the heuristic search algorithm according to the number of nodes of the discrete temperature control sequence, and initializing parameter settings of the heuristic search algorithm; The initial population is randomly generated, each individual corresponds to a candidate control subsequence, and the gene of the individual is composed of the temperature setting value and duration of the temperature control node; For each individual in the initial population, the candidate control subsequence is input into the finite element model of the thermal shock testing machine, and the three-dimensional temperature distribution change process of the test area under the candidate control subsequence is simulated to obtain the objective function value reflecting the temperature uniformity and the constraint violation data reflecting the control accuracy; According to the objective function value and the constraint violation data, a weighted sum method is used to calculate the fitness value of each individual, and the higher the fitness value, the better the optimization performance of the candidate control subsequence corresponding to the individual; The individuals in the population are selected, crossover and mutated to generate a new generation of population until the preset number of iterations is reached or the convergence condition is met, and multiple temperature control subsequences are obtained.

7. A thermal shock test machine control device, used to execute the thermal shock test machine control method according to any one of claims 1 to 6, characterized in that: The device comprises: An acquisition module is used to acquire the physical structure and material properties of the hot and cold shock testing machine, and to construct a temperature prediction model of the hot and cold shock testing machine; specifically, the module comprises: performing three-dimensional modeling on the physical structure of the hot and cold shock testing machine to obtain an initial three-dimensional geometric model, and setting material parameters of the three-dimensional geometric model according to the material properties of the hot and cold shock testing machine to obtain a target three-dimensional physical model; performing meshing on the target three-dimensional physical model to obtain a finite element model composed of a plurality of tetrahedral units; according to the finite element model, establishing a three-dimensional Fourier equation describing the transient heat transfer process inside the hot and cold shock testing machine, and combining boundary conditions and initial conditions to obtain a group of partial differential equations for heat conduction; and performing meshing on the target three-dimensional physical model to obtain a finite element model composed of a plurality of tetrahedral units. The conduction partial differential equations are spatially discretized to obtain the heat conduction ordinary differential equations, and the heat conduction ordinary differential equations are time discretized to obtain the heat conduction algebraic equations; the heat conduction algebraic equations are solved to obtain the temperature response sequence of the hot and cold shock tester, and based on the temperature response sequence, a high-dimensional temperature response function describing the dynamic temperature distribution of the hot and cold shock tester is constructed; the high-dimensional temperature response function is approximated by dimensionality reduction using a long short-term memory neural network to obtain a low-dimensional prediction model, and the low-dimensional prediction model is simulated and verified and error analyzed to obtain an error analysis result; according to the error analysis result, the low-dimensional prediction model is iteratively optimized in parameters to obtain a temperature prediction model; A fusion module, used to arrange multiple temperature sensors in the test area and obtain multiple temperature sensor data, and perform adaptive weighted fusion on the multiple temperature sensor data to generate three-dimensional temperature distribution data of the test area; A solution module, used for constructing a model prediction controller according to the temperature prediction model and the three-dimensional temperature distribution data, and solving the optimal control problem through the model prediction controller to generate a first temperature control sequence; Establishing a module for acquiring the operating status data of the refrigeration system and establishing a corresponding thermodynamic model, and generating an adaptive refrigeration strategy for the refrigeration system according to the thermodynamic model; A generating module is used to perform control optimization on the first temperature control sequence according to the adaptive refrigeration strategy to generate a second temperature control sequence.

8. A computer device, characterized in that: The computer device comprises: a memory and at least one processor, wherein instructions are stored in the memory; The at least one processor calls the instructions in the memory to enable the computer device to execute the thermal shock tester control method according to any one of claims 1 to 6.

9. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions are executed by the processor, the hot and cold shock testing machine control method according to any one of claims 1 to 6 is implemented.

Images