Electromechanical system simulation parameter single working condition calibration method

Abstract

The invention provides an electromechanical system simulation parameter single working condition calibration method. The method comprises the following steps: constructing a target function by taking the highest consistency of simulation data output by an electromechanical system simulation model under a target working condition and actual measurement data output by an actual electromechanical system as a target; performing single-objective optimization on the objective function, and adding a single-objective optimization result into an experience knowledge base; dividing each solution in the experience knowledge base into a plurality of subspaces so as to carry out primary Bayesian optimization in each subspace; adding the primary Bayesian optimization result into an experience knowledge base, and then carrying out multi-modal detection on each solution to carry out secondary Bayesian optimization in each modal; and adding the secondary Bayesian optimization result into an experience knowledge base, and performing circulation again based on the secondary Bayesian optimization result to obtain a single-target optimization result when a termination condition is met as a simulation parameter single-working-condition calibration result of the electromechanical system simulation model. According to the method, the real optimal solution of the simulation parameters of the electromechanical system can be obtained more accurately.

Description

Technical Field

[0001] The present invention relates to the technical field of simulation model calibration, and in particular to a single-condition calibration method for electromechanical system simulation parameters. Background Art

[0002] Electromechanical systems are widely used in manufacturing, automotive, aerospace, medical equipment, consumer electronics, energy systems and other fields. They are the cross-integration of mechanical engineering, electronic engineering, control engineering, computer science and information technology, and are highly integrated. Due to the high integration of electromechanical systems, most of the functional and performance tests need to be conducted with the help of simulation models. Therefore, it is necessary to establish an accurate electromechanical system simulation model.

[0003] The essence of the electromechanical system simulation model is the approximation of the actual electromechanical system. In order to establish an accurate electromechanical system simulation model, it is usually necessary to first evaluate the consistency between the simulation data output by the electromechanical system simulation model and the measured data output by the actual electromechanical system. When the consistency is poor, the simulation parameters of the electromechanical system simulation model are adjusted to regain the simulation data, and then the consistency is evaluated again until the simulation data and the measured data achieve a high degree of consistency. This process is called the simulation parameter calibration of the electromechanical system simulation model.

[0004] In the process of calibrating the simulation parameters of the electromechanical system simulation model, the inventor found that: currently, the simulation parameter calibration of the electromechanical system simulation model is usually realized by using an evolutionary algorithm assisted by a multi-agent model, so as to overcome the problem that the objective function of the electromechanical system simulation model has a high computational cost and the simulation model is expressed as a complex black box function, and efficiently search for the global optimal solution of the electromechanical system simulation model under limited computing resources. However, even if the electromechanical system simulation parameters are calibrated under a single working condition, since the consistency between the simulation data and the measured data is obtained after integrating multiple indicators of the electromechanical system, the calibration of the electromechanical system simulation parameters under a single working condition may also have multiple global optimal solutions or multiple global or local optimal solutions, and decision makers often need to find multiple global optimal solutions or the true optimal solution among multiple global or local optimal solutions. Summary of the invention

[0005] The embodiment of the present invention provides a single-operating-condition calibration method for electromechanical system simulation parameters, so as to solve the problem that the current single-operating-condition calibration method for electromechanical system simulation parameters is often difficult to accurately determine the true optimal solution.

[0006] In a first aspect, an embodiment of the present invention provides a single-condition calibration method for electromechanical system simulation parameters, comprising:

[0007] The objective function is constructed with the goal of achieving the highest consistency between the simulation data output by the electromechanical system simulation model under the target working condition and the measured data output by the actual electromechanical system;

[0008] Taking each simulation parameter of the electromechanical system simulation model as a decision vector, single-objective optimization is performed on the objective function, and the single-objective optimization result is added to the experience knowledge base;

[0009] Dividing each solution of each simulation parameter of the electromechanical system simulation model in the empirical knowledge base into a plurality of subspaces, and constructing a proxy model in each subspace for initial Bayesian optimization;

[0010] The initial Bayesian optimization result is added to the empirical knowledge base, multi-modal detection is performed on each solution in the empirical knowledge base, and a proxy model is constructed in each mode obtained by the detection to perform secondary Bayesian optimization;

[0011] The quadratic Bayesian optimization result is added to the empirical knowledge base, and the step of "performing single objective optimization on the objective function" and subsequent steps are re-executed based on the quadratic Bayesian optimization result until the termination condition is met, and the single working condition calibration result of the simulation parameters of the electromechanical system simulation model is obtained according to the single objective optimization result.

[0012] In a possible implementation, each solution of each simulation parameter of the electromechanical system simulation model in the empirical knowledge base is divided into a plurality of subspaces, including:

[0013] Obtaining the optimal solution from among the solutions for each simulation parameter of the electromechanical system simulation model in the experience knowledge base;

[0014] Taking the optimal solution as the center, determine the (n-1) solutions closest to the optimal solution to form a current subspace with n solutions;

[0015] The current subspace is deleted from the empirical knowledge base to obtain a new empirical knowledge base, and the step of "obtaining the optimal solution among the solutions for each simulation parameter of the electromechanical system simulation model in the empirical knowledge base" and subsequent steps are re-executed until the number of solutions in the new empirical knowledge base is less than n, and multiple subspaces are obtained.

[0016] In a possible implementation, the process of determining the number of subspaces is as follows:

[0017] according to determining the number of the subspaces;

[0018] Among them, m(N t ) is the number of subspaces at iteration time t, N t is the number of solutions in the empirical knowledge base at iteration time t, n is the number of solutions in the subspace, θ is the subspace quantity control parameter, θ∈[0.5,1], N max The maximum number of objective function calculations of the algorithm.

[0019] In a possible implementation, after obtaining the multiple subspaces, the method further includes:

[0020] Each subspace is recorded as the first subspace, and each edge solution in each first subspace is taken as the center, and the (n-1) solutions closest to each edge solution are determined to form multiple new subspaces with n solutions, which are recorded as the second subspace;

[0021] The constructing of a proxy model in each subspace for initial Bayesian optimization includes:

[0022] Determine each target subspace according to each of the first subspaces and each of the second subspaces;

[0023] A proxy model is constructed in each target subspace to perform initial Bayesian optimization.

[0024] In a possible implementation manner, determining each target subspace according to each of the first subspaces and each of the second subspaces includes:

[0025] Calculate the union of each second subspace corresponding to each of the first subspaces, and record it as the second target subspace corresponding to each of the first subspaces;

[0026] Each of the first subspaces and each of the second target subspaces are respectively determined as a target subspace.

[0027] In a possible implementation, multimodal detection is performed on each solution in the experience knowledge base, including:

[0028] Select the best solution from each solution in the experience knowledge base as a seed;

[0029] respectively calculating the distance between each solution in the experience knowledge base except each of the seeds and each of the seeds;

[0030] Compare the distance between each solution in the experience knowledge base except each of the seeds and each of the seeds with a set distance threshold respectively;

[0031] If the distance between a solution in the experience knowledge base and each of the seeds is greater than the set distance threshold, the solution is used as a new seed;

[0032] Based on the new seeds, re-execute the step of "respectively calculating the distance between each solution in the empirical knowledge base except each of the seeds and each of the seeds" and subsequent steps until all solutions in the empirical knowledge base are traversed to obtain multiple seeds in the empirical knowledge base, wherein each seed represents a mode.

[0033] In a possible implementation, the process of determining the set distance threshold is:

[0034] Calculate the distance between the optimal solution in each subspace and other solutions in the subspace, recorded as the first distance;

[0035] Determine a representative distance corresponding to each subspace according to each of the first distances in each subspace;

[0036] The set distance threshold is determined according to the representative distance corresponding to each subspace.

[0037] In a second aspect, an embodiment of the present invention provides a single-condition calibration device for electromechanical system simulation parameters, comprising:

[0038] An objective function construction module is used to construct an objective function with the goal of achieving the highest consistency between the simulation data output by the electromechanical system simulation model under the target working condition and the measured data output by the actual electromechanical system;

[0039] A first optimization module is used to use each simulation parameter of the electromechanical system simulation model as a decision vector, perform single-objective optimization on the objective function, and add the single-objective optimization result to the experience knowledge base;

[0040] A second optimization module is used to divide each solution of each simulation parameter of the electromechanical system simulation model in the empirical knowledge base into multiple subspaces, and construct a proxy model in each subspace to perform initial Bayesian optimization;

[0041] A third optimization module is used to add the initial Bayesian optimization result to the empirical knowledge base, perform multimodal detection on each solution in the empirical knowledge base, and construct a proxy model in each mode obtained by the detection to perform secondary Bayesian optimization;

[0042] A cyclic calibration module is used to add the quadratic Bayesian optimization results to the empirical knowledge base, and re-execute the step of "performing single-objective optimization on the objective function" and subsequent steps based on the quadratic Bayesian optimization results until the termination condition is met, and obtain the single-condition calibration results of the simulation parameters of the electromechanical system simulation model according to the single-objective optimization results.

[0043] In a third aspect, an embodiment of the present invention provides a terminal, comprising a memory and a processor, wherein the memory is used to store a computer program, and the processor is used to call and run the computer program stored in the memory to perform the steps of the method described in the first aspect or any possible implementation of the first aspect.

[0044] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method described in the first aspect or any possible implementation method of the first aspect are implemented.

[0045] The embodiment of the present invention provides a single-condition calibration method for electromechanical system simulation parameters, which constructs an objective function with the highest consistency between simulation data output by an electromechanical system simulation model under a target condition and measured data output by an actual electromechanical system as the goal; then, each simulation parameter of the electromechanical system simulation model is used as a decision vector, the objective function is single-objective optimized, and the single-objective optimization result is added to an empirical knowledge base; then, each solution of each simulation parameter of the electromechanical system simulation model in the empirical knowledge base is divided into multiple subspaces, and a proxy model is constructed in each subspace for initial Bayesian optimization; then, the initial Bayesian optimization result is added to the empirical knowledge base, multi-modal detection is performed on each solution in the empirical knowledge base, and a proxy model is constructed in each mode obtained by the detection for secondary Bayesian optimization; finally, the secondary Bayesian optimization result is added to the empirical knowledge base, and based on the secondary Bayesian optimization result, the cycle of "single-objective optimization of the objective function" is re-executed, and then the single-condition calibration result of the simulation parameters of the electromechanical system simulation model is obtained according to the single-objective optimization result when the termination condition is met. On the one hand, the global optimization of the simulation parameters of the electromechanical system simulation model can be achieved through single-objective optimization. On the other hand, the initial Bayesian optimization based on the global optimization can be used to perform more targeted local optimization. Moreover, after obtaining the local optimization of the initial Bayesian optimization, the local optimization of each mode (that is, each global optimal solution or local optimal solution) in the result of the local optimization is performed again through the secondary Bayesian optimization, which can more accurately obtain the true optimal solution of the electromechanical system simulation parameters, thereby helping to improve the accuracy of the electromechanical system simulation model under a single working condition. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0047] Figure 1 It is a flow chart for implementing a single-condition calibration method for electromechanical system simulation parameters provided by an embodiment of the present invention;

[0048] Figure 2 It is a block diagram of an implementation method of a single-condition calibration method for electromechanical system simulation parameters provided by an embodiment of the present invention;

[0049] Figure 3 It is a structural schematic diagram of a single-condition calibration device for electromechanical system simulation parameters provided by an embodiment of the present invention;

[0050] Figure 4 is a schematic diagram of a terminal provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0051] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present invention. However, it should be clear to those skilled in the art that the present invention may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present invention.

[0052] In order to make the purpose, technical solutions and advantages of the present invention more clear, specific embodiments will be described below in conjunction with the accompanying drawings.

[0053] See also Figure 1 , which shows a flowchart of the implementation of the electromechanical system simulation parameter single-condition calibration method provided by an embodiment of the present invention, and is described in detail as follows:

[0054] In step 101 , an objective function is constructed with the goal of achieving the highest consistency between the simulation data output by the electromechanical system simulation model under the target working condition and the measured data output by the actual electromechanical system.

[0055] For example, the objective function can be to maximize Among them, ω i is the weight coefficient of the i-th indicator under the target working condition, s i is the consistency evaluation score of the i-th indicator under the target condition.

[0056] The objective function is constructed with the goal of maximizing the consistency between the simulation data output by the electromechanical system simulation model under the target working conditions and the measured data output by the actual electromechanical system. The minimum error of different indicators can be converted into the highest consistency of each indicator, thereby converting the multi-objective optimization problem into a single-objective optimization problem, so as to facilitate the optimal calibration of the simulation parameters of the electromechanical system simulation model.

[0057] In step 102, each simulation parameter of the electromechanical system simulation model is used as a decision vector to perform single-objective optimization on the objective function, and the single-objective optimization result is added to the experience knowledge base.

[0058] Exemplary, combined Figure 2As shown in the figure, the particle swarm optimization (PSO) algorithm can be used to optimize the objective function. PSO imitates the social behavior of a flock of birds looking for food, where each "particle" represents a potential solution in the solution space. Particles update their positions based on the experience of individuals and groups, effectively exploring the solution space. This algorithm is simple, intuitive and easy to implement, and is particularly suitable for complex optimization scenarios where the analytical model is unclear or there is no gradient information.

[0059] Among them, when performing single-objective optimization based on PSO, initialization is first performed to initialize the population size, the speed and position of each particle.

[0060] Then, the fitness value of the initial population, that is, the value of the objective function corresponding to the particle, is calculated, and the particle's historical optimal solution and the global optimal solution are recorded.

[0061] Then, based on the initialized particle speed and position, as well as the particle's historical optimal solution and the global optimal solution, the particle speed and position are updated by combining the speed update formula and the position update formula.

[0062] Then, the fitness value corresponding to the updated particle is calculated, and the historical optimal solution and the global optimal solution of the particle can be updated according to the fitness value corresponding to the updated particle.

[0063] In this embodiment, the PSO algorithm is used to simulate the social behavior of bird flocks to achieve effective global search. Its group wisdom helps avoid falling into local optimality and improves the probability of finding the global optimal solution. In addition, the historical information recorded by the PSO algorithm (that is, the historical global optimal solution) can be added to the experience knowledge base to provide dynamic prior knowledge for subsequent Bayesian optimization through the experience knowledge base, further enhancing the breadth and depth of the search.

[0064] In step 103, each solution of each simulation parameter of the electromechanical system simulation model in the empirical knowledge base is divided into multiple subspaces, and a proxy model is constructed in each subspace to perform initial Bayesian optimization.

[0065] Among them, Bayesian Optimization (BO) uses Gaussian process as a proxy model, combining global search and local optimization to deal with unknown objective functions. By integrating prior knowledge and real-time data updates, the algorithm predicts the behavior of the objective function with uncertainty. BO is particularly suitable for complex optimization problems that need to consider the impact of measurement noise.

[0066] In this embodiment, taking into account the high computational cost and long update cycle of BO when processing large-scale data sets or high-dimensional problems, each solution in the empirical knowledge base is divided into multiple subspaces, and a proxy model is constructed in each subspace for initial Bayesian optimization. By constructing a high-precision proxy model through Bayesian optimization of multiple subspaces, fast and accurate local optimization can be achieved with limited computing resources.

[0067] In step 104, the initial Bayesian optimization result is added to the empirical knowledge base, multimodal detection is performed on each solution in the empirical knowledge base, and a proxy model is constructed in each mode obtained by the detection to perform secondary Bayesian optimization.

[0068] Combination Figure 2 In this embodiment, after adding the initial Bayesian optimization result to the empirical knowledge base, considering that the solution in the empirical knowledge base may contain multiple global optimal solutions or multiple global optimal solutions and multiple local optimal solutions, in order to more accurately determine the true optimal solution, multimodal detection is performed on each solution in the empirical knowledge base, that is, how many global optimal solutions and / or local optimal solutions are specifically contained in the empirical knowledge base is detected, so as to further search for the optimal solution within each global optimal solution and / or local optimal solution.

[0069] In step 105, the quadratic Bayesian optimization result is added to the empirical knowledge base, and the step of "performing single objective optimization on the objective function" and subsequent steps are re-executed based on the quadratic Bayesian optimization result until the termination condition is met, and the single working condition calibration result of the simulation parameters of the electromechanical system simulation model is obtained according to the single objective optimization result.

[0070] like Figure 2 As shown, in this embodiment, in each iteration, the optimal solution of PSO is fed back to the primary Bayesian optimization, the optimal solution of the primary Bayesian optimization is fed back to the secondary Bayesian optimization, and the optimal solution of the secondary Bayesian optimization is fed back to PSO, thereby realizing dynamic information sharing in the optimization process, which can not only improve the global search capability, but also significantly improve the overall optimization performance.

[0071] The embodiment of the present invention constructs an objective function by taking the highest consistency between the simulation data output by the electromechanical system simulation model under the target working condition and the measured data output by the actual electromechanical system as the goal; then, each simulation parameter of the electromechanical system simulation model is used as a decision vector to perform single-objective optimization on the objective function, and the single-objective optimization result is added to the empirical knowledge base; then, each solution of each simulation parameter of the electromechanical system simulation model in the empirical knowledge base is divided into multiple subspaces, and a proxy model is constructed in each subspace for initial Bayesian optimization; then, the initial Bayesian optimization result is added to the empirical knowledge base, multi-modal detection is performed on each solution in the empirical knowledge base, and a proxy model is constructed in each mode obtained by the detection for secondary Bayesian optimization; finally, the secondary Bayesian optimization result is added to the empirical knowledge base, and based on the secondary Bayesian optimization result, the cycle of "single-objective optimization of the objective function" is re-executed, and then the simulation parameter single-condition calibration result of the electromechanical system simulation model is obtained according to the single-objective optimization result when the termination condition is met. On the one hand, the global optimization of the simulation parameters of the electromechanical system simulation model can be achieved through single-objective optimization. On the other hand, the initial Bayesian optimization based on the global optimization can be used to perform more targeted local optimization. Moreover, after obtaining the local optimization of the initial Bayesian optimization, the local optimization of each mode (that is, each global optimal solution or local optimal solution) in the result of the local optimization is performed again through the secondary Bayesian optimization, which can more accurately obtain the true optimal solution of the electromechanical system simulation parameters, thereby helping to improve the accuracy of the electromechanical system simulation model under a single working condition.

[0072] In one embodiment, dividing each solution of each simulation parameter of the electromechanical system simulation model in the empirical knowledge base into multiple subspaces may include:

[0073] The optimal solution among various solutions of each simulation parameter of the electromechanical system simulation model in the experience knowledge base is obtained.

[0074] Taking the optimal solution as the center, determine the (n-1) solutions closest to the optimal solution to form the current subspace with n solutions.

[0075] The current subspace is deleted from the empirical knowledge base to obtain a new empirical knowledge base, and the step of "obtaining the optimal solution among the solutions for each simulation parameter of the electromechanical system simulation model in the empirical knowledge base" and subsequent steps are re-executed until the number of solutions in the new empirical knowledge base is less than n, and multiple subspaces are obtained.

[0076] For example, assuming that the empirical knowledge base D is known, the number of samples in each subspace (that is, the number of solutions) is n. First, the optimal sample point (that is, the optimal solution) can be selected from the empirical knowledge base D as the starting point. This can be obtained by α1=argminf(x), x∈K1, where K1=D is the current empirical knowledge base.

[0077] In the process of constructing the subspace, α1 is taken as the center, and the n-1 sample points closest to it are found to form the subspace set M1. Specifically, the n-1 sample points closest to α1 can be found by using a distance metric such as Euclidean distance.

[0078] Then, the subspace construction is repeated, that is, the samples contained in M1 are removed from K1 or D, that is, K2=K1\M1, and then the above process of finding the optimal point and constructing the subspace is repeated in K2 until all samples are assigned to a subspace, and finally a subspace set is obtained. Each subspace can focus on the optimization of its specific area to enhance the global search capability.

[0079] In this embodiment, by dividing a large data set into multiple smaller subsets and building proxy models in parallel on these subsets, the overall computational burden can be significantly reduced and the optimization process can be accelerated. In addition, the existing sample information can be used more accurately and efficiently, providing the possibility of dynamic adjustment for the optimization algorithm.

[0080] Exemplarily, the process of determining the number of subspaces may be:

[0081] according to Determine the number of subspaces.

[0082] Among them, m(N t ) is the number of subspaces at iteration time t, N t is the number of solutions in the empirical knowledge base at iteration time t, n is the number of solutions in the subspace, θ is the subspace number control parameter, θ∈[0.5,1], N max The maximum number of objective function calculations of the algorithm.

[0083] This embodiment considers that as the number of samples in the experience knowledge base continues to increase, the number of subspaces also increases accordingly. In order to provide appropriate exploration and convergence speed at different stages of the algorithm, the number of subspaces m (N t ), in the early stage of algorithm iteration, a larger number of subspaces is used to provide extensive space exploration to avoid falling into the local optimum. In the later stage of algorithm iteration, a smaller number of subspaces is used to speed up the convergence speed and focus on the subspaces with higher confidence. Moreover, the calculation formula for determining the number of subspaces is simple, which can avoid introducing complex calculation processes.

[0084] In one embodiment, after obtaining the multiple subspaces, the following steps may also be performed:

[0085] Each subspace is recorded as the first subspace. Taking each edge solution in each first subspace as the center, the (n-1) solutions closest to each edge solution are determined to form multiple new subspaces with n solutions, which are recorded as the second subspace.

[0086] Accordingly, constructing a proxy model in each subspace for initial Bayesian optimization may include:

[0087] According to each of the first subspaces and each of the second subspaces, each of the target subspaces is determined.

[0088] A surrogate model is constructed in each target subspace for initial Bayesian optimization.

[0089] In this embodiment, after the empirical knowledge base is divided into multiple subspaces, in order to enhance the exploration of the edges of each subspace, multiple new subspaces are constructed with each edge solution in each subspace as the center, so as to fully explore local solutions and edge solutions through the subspaces divided by the empirical knowledge base and the new subspaces, thereby helping to optimize the overall optimization performance.

[0090] Exemplarily, determining each target subspace according to each first subspace and each second subspace may include:

[0091] The union of each second subspace corresponding to each first subspace is calculated, and recorded as the second target subspace corresponding to each first subspace.

[0092] Each first subspace and each second target subspace are respectively determined as a target subspace.

[0093] In this embodiment, in order to fully explore the edges of each subspace and avoid adding too many subspaces, after constructing each second subspace corresponding to each first subspace, consider calculating the union of each second subspace as the second target subspace, so as to fully explore the edges of each subspace using the second target subspace.

[0094] In one embodiment, performing multimodal detection on each solution in the experience knowledge base may include:

[0095] The best solution is selected from various solutions in the empirical knowledge base as a seed.

[0096] The distance between each solution in the empirical knowledge base except each seed and each seed is calculated respectively.

[0097] The distance between each solution except each seed in the experience knowledge base and each seed is compared with the set distance threshold.

[0098] If the distance between a solution in the experience knowledge base and each seed is greater than the set distance threshold, the solution will be used as a new seed.

[0099] Based on the new seeds, re-execute the steps of "respectively calculating the distance between each solution in the empirical knowledge base except each seed and each seed" and subsequent steps until all solutions in the empirical knowledge base are traversed to obtain multiple seeds in the empirical knowledge base, where each seed represents a mode.

[0100] Exemplarily, the process of determining the distance threshold may be:

[0101] Calculate the distance between the optimal solution in each subspace and other solutions in the subspace, and record it as the first distance.

[0102] According to each first distance in each subspace, a representative distance corresponding to each subspace is determined.

[0103] According to the representative distance corresponding to each subspace, a distance threshold is determined.

[0104] In this embodiment, when performing multimodal detection on each solution in the empirical knowledge base, in order to more reasonably identify different modes, the distance threshold can be determined by using the distance between the optimal solution and other solutions in each subspace when the subspace is divided before the initial Bayesian optimization. For example, the representative distance of each subspace is determined by the mean of each first distance in each subspace, and the distance threshold is determined by the mean of the representative distance of each subspace. Alternatively, the representative distance of each subspace is determined by the median and maximum values ​​of each first distance in each subspace, and the distance threshold is determined by the mean median and maximum values ​​of the representative distance of each subspace.

[0105] It should be understood that the order of execution of the steps in the above embodiment does not necessarily mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present invention.

[0106] The following is an embodiment of the device of the present invention. For details not described in detail therein, reference may be made to the corresponding method embodiment described above.

[0107] Figure 3 The schematic diagram of the structure of the electromechanical system simulation parameter single-condition calibration device provided by the embodiment of the present invention is shown. For the convenience of explanation, only the part related to the embodiment of the present invention is shown, which is described in detail as follows:

[0108] like Figure 3 As shown, the single-condition calibration device for electromechanical system simulation parameters includes: an objective function construction module 31, a first optimization module 32, a second optimization module 33, a third optimization module 34 and a cyclic calibration module 35.

[0109] An objective function construction module 31 is used to construct an objective function with the goal of achieving the highest consistency between the simulation data output by the electromechanical system simulation model under the target working condition and the measured data output by the actual electromechanical system;

[0110] A first optimization module 32 is used to use each simulation parameter of the electromechanical system simulation model as a decision vector, perform single-objective optimization on the objective function, and add the single-objective optimization result to the experience knowledge base;

[0111] A second optimization module 33 is used to divide each solution of each simulation parameter of the electromechanical system simulation model in the empirical knowledge base into multiple subspaces, and construct a proxy model in each subspace to perform initial Bayesian optimization;

[0112] The third optimization module 34 is used to add the initial Bayesian optimization result to the empirical knowledge base, perform multimodal detection on each solution in the empirical knowledge base, and construct a proxy model in each mode obtained by the detection to perform secondary Bayesian optimization;

[0113] The cyclic calibration module 35 is used to add the quadratic Bayesian optimization results to the empirical knowledge base, and re-execute the step of "performing single-objective optimization on the objective function" and subsequent steps based on the quadratic Bayesian optimization results until the termination condition is met, and obtain the single-condition calibration results of the simulation parameters of the electromechanical system simulation model according to the single-objective optimization results.

[0114] The embodiment of the present invention constructs an objective function by taking the highest consistency between the simulation data output by the electromechanical system simulation model under the target working condition and the measured data output by the actual electromechanical system as the goal; then, each simulation parameter of the electromechanical system simulation model is used as a decision vector to perform single-objective optimization on the objective function, and the single-objective optimization result is added to the empirical knowledge base; then, each solution of each simulation parameter of the electromechanical system simulation model in the empirical knowledge base is divided into multiple subspaces, and a proxy model is constructed in each subspace for initial Bayesian optimization; then, the initial Bayesian optimization result is added to the empirical knowledge base, multi-modal detection is performed on each solution in the empirical knowledge base, and a proxy model is constructed in each mode obtained by the detection for secondary Bayesian optimization; finally, the secondary Bayesian optimization result is added to the empirical knowledge base, and based on the secondary Bayesian optimization result, the cycle of "single-objective optimization of the objective function" is re-executed, and then the simulation parameter single-condition calibration result of the electromechanical system simulation model is obtained according to the single-objective optimization result when the termination condition is met. On the one hand, the global optimization of the simulation parameters of the electromechanical system simulation model can be achieved through single-objective optimization. On the other hand, the initial Bayesian optimization based on the global optimization can be used to perform more targeted local optimization. Moreover, after obtaining the local optimization of the initial Bayesian optimization, the local optimization of each mode (that is, each global optimal solution or local optimal solution) in the result of the local optimization is performed again through the secondary Bayesian optimization, which can more accurately obtain the true optimal solution of the electromechanical system simulation parameters, thereby helping to improve the accuracy of the electromechanical system simulation model under a single working condition.

[0115] In one possible implementation, the second optimization module 33 can be used to obtain the optimal solution among the solutions for each simulation parameter of the electromechanical system simulation model in the empirical knowledge base; with the optimal solution as the center, determine the (n-1) solutions closest to the optimal solution to form a current subspace with n solutions; delete the current subspace from the empirical knowledge base to obtain a new empirical knowledge base, and re-execute the step of "obtaining the optimal solution among the solutions for each simulation parameter of the electromechanical system simulation model in the empirical knowledge base" and subsequent steps until the number of solutions in the new empirical knowledge base is less than n, and multiple subspaces are obtained.

[0116] In a possible implementation, the second optimization module 33 may be used to determining the number of the subspaces;

[0117] Among them, m(N t ) is the number of subspaces at iteration time t, N t is the number of solutions in the empirical knowledge base at iteration time t, n is the number of solutions in the subspace, θ is the subspace quantity control parameter, θ∈[0.5,1], N maxThe maximum number of objective function calculations of the algorithm.

[0118] In one possible implementation, the second optimization module 33 can also be used to record each subspace as a first subspace, and determine the (n-1) solutions closest to each edge solution in each first subspace as the center, to form multiple new subspaces with n solutions, which are recorded as second subspaces; constructing a proxy model in each subspace for initial Bayesian optimization, including: determining each target subspace based on each of the first subspaces and each of the second subspaces; and constructing a proxy model in each of the target subspaces for initial Bayesian optimization.

[0119] In a possible implementation, the second optimization module 33 can be used to calculate the union of each second subspace corresponding to each first subspace, recorded as the second target subspace corresponding to each first subspace; and determine each first subspace and each second target subspace as a target subspace respectively.

[0120] In a possible implementation, the third optimization module 34 can be used to select the best solution from the solutions in the experience knowledge base as a seed; respectively calculate the distance between each solution in the experience knowledge base except each seed and each seed; respectively compare the distance between each solution in the experience knowledge base except each seed and each seed with a set distance threshold; if the distance between a solution in the experience knowledge base and each seed is greater than the set distance threshold, then use the solution as a new seed; based on the new seed, re-execute "respectively calculate the distance between each solution in the experience knowledge base except each seed and each seed" and subsequent steps until all solutions in the experience knowledge base are traversed to obtain multiple seeds in the experience knowledge base, wherein each seed represents a mode.

[0121] In one possible implementation, the third optimization module 34 can be used to calculate the distance between the optimal solution in each subspace and other solutions in the subspace, recorded as the first distance; determine the representative distance corresponding to each subspace based on the first distances in each subspace; and determine the set distance threshold based on the representative distance corresponding to each subspace.

[0122] Figure 4 is a schematic diagram of a terminal provided by an embodiment of the present invention. Figure 4 As shown, the terminal 4 of this embodiment includes: a processor 40, a memory 41, and a computer program 42 stored in the memory 41 and executable on the processor 40. When the processor 40 executes the computer program 42, the steps in the above-mentioned electromechanical system simulation parameter single-condition calibration method embodiments are implemented, for example Figure 1Alternatively, when the processor 40 executes the computer program 42, the functions of each module / unit in the above-mentioned device embodiments are realized, for example Figure 3 Functionality of the modules / units 31 to 35 shown.

[0123] Exemplarily, the computer program 42 may be divided into one or more modules / units, one or more modules / units are stored in the memory 41 and executed by the processor 40 to implement the present invention. One or more modules / units may be a series of computer program instruction segments capable of implementing specific functions, and the instruction segments are used to describe the execution process of the computer program 42 in the terminal 4. For example, the computer program 42 may be divided into Figure 3 Modules / units 31 to 35 are shown.

[0124] The terminal 4 may be a computing device such as a desktop computer, a notebook, a PDA, or a cloud server. The terminal 4 may include, but is not limited to, a processor 40 and a memory 41. Those skilled in the art will appreciate that Figure 4 It is only an example of terminal 4 and does not constitute a limitation on terminal 4. It may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, the terminal may also include input and output devices, network access devices, buses, etc.

[0125] The processor 40 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0126] The memory 41 may be an internal storage unit of the terminal 4, such as a hard disk or memory of the terminal 4. The memory 41 may also be an external storage device of the terminal 4, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the terminal 4. Further, the memory 41 may also include both an internal storage unit of the terminal 4 and an external storage device. The memory 41 is used to store computer programs and other programs and data required by the terminal. The memory 41 may also be used to temporarily store data that has been output or is to be output.

[0127] The technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0128] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0129] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0130] In the embodiments provided by the present invention, it should be understood that the disclosed devices / terminals and methods can be implemented in other ways. For example, the device / terminal embodiments described above are only schematic, for example, the division of modules or units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0131] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, the functional units in the various embodiments of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated units may be implemented in the form of hardware or in the form of software functional units.

[0132] If the integrated module / 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 present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, the steps of the above-mentioned electromechanical system simulation parameter single-condition calibration method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal and software distribution medium, etc.

[0133] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such 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 invention, and should be included in the protection scope of the present invention.

Claims

1. A single-condition calibration method for electromechanical system simulation parameters, characterized in that: include: The objective function is constructed with the goal of achieving the highest consistency between the simulation data output by the electromechanical system simulation model under the target working condition and the measured data output by the actual electromechanical system; Taking each simulation parameter of the electromechanical system simulation model as a decision vector, single-objective optimization is performed on the objective function, and the single-objective optimization result is added to the experience knowledge base; Dividing each solution of each simulation parameter of the electromechanical system simulation model in the empirical knowledge base into a plurality of subspaces, and constructing a proxy model in each subspace for initial Bayesian optimization; The initial Bayesian optimization result is added to the empirical knowledge base, multi-modal detection is performed on each solution in the empirical knowledge base, and a proxy model is constructed in each mode obtained by the detection to perform secondary Bayesian optimization; The quadratic Bayesian optimization result is added to the empirical knowledge base, and the step of "performing single objective optimization on the objective function" and subsequent steps are re-executed based on the quadratic Bayesian optimization result until the termination condition is met, and the single working condition calibration result of the simulation parameters of the electromechanical system simulation model is obtained according to the single objective optimization result.

2. The electromechanical system simulation parameter single-condition calibration method according to claim 1, characterized in that: Dividing each solution of each simulation parameter of the electromechanical system simulation model in the empirical knowledge base into a plurality of subspaces, including: Obtaining the optimal solution from among the solutions for each simulation parameter of the electromechanical system simulation model in the experience knowledge base; Taking the optimal solution as the center, determine the (n-1) solutions closest to the optimal solution to form a current subspace with n solutions; The current subspace is deleted from the empirical knowledge base to obtain a new empirical knowledge base, and the step of "obtaining the optimal solution among the solutions for each simulation parameter of the electromechanical system simulation model in the empirical knowledge base" and subsequent steps are re-executed until the number of solutions in the new empirical knowledge base is less than n, and multiple subspaces are obtained.

3. The electromechanical system simulation parameter single-condition calibration method according to claim 2, characterized in that: The process of determining the number of subspaces is as follows: according to determining the number of the subspaces; Among them, m(N t ) is the number of subspaces at iteration time t, N t is the number of solutions in the empirical knowledge base at iteration time t, n is the number of solutions in the subspace, θ is the subspace quantity control parameter, θ∈[0.5,1], N max The maximum number of objective function calculations of the algorithm.

4. The electromechanical system simulation parameter single-condition calibration method according to claim 2, characterized in that: After obtaining multiple subspaces, it also includes: Each subspace is recorded as the first subspace, and each edge solution in each first subspace is taken as the center, and the (n-1) solutions closest to each edge solution are determined to form multiple new subspaces with n solutions, which are recorded as the second subspace; The constructing of a proxy model in each subspace for initial Bayesian optimization includes: Determine each target subspace according to each of the first subspaces and each of the second subspaces; A proxy model is constructed in each target subspace to perform initial Bayesian optimization.

5. The electromechanical system simulation parameter single-condition calibration method according to claim 4, characterized in that: Determining each target subspace according to each of the first subspaces and each of the second subspaces includes: Calculate the union of each second subspace corresponding to each of the first subspaces, and record it as the second target subspace corresponding to each of the first subspaces; Each of the first subspaces and each of the second target subspaces are respectively determined as a target subspace.

6. The electromechanical system simulation parameter single-condition calibration method according to claim 1, characterized in that: Multimodal detection is performed on each solution in the empirical knowledge base, including: Select the best solution from each solution in the experience knowledge base as a seed; respectively calculating the distance between each solution in the experience knowledge base except each of the seeds and each of the seeds; Compare the distance between each solution in the experience knowledge base except each of the seeds and each of the seeds with a set distance threshold respectively; If the distance between a solution in the experience knowledge base and each of the seeds is greater than the set distance threshold, the solution is used as a new seed; Based on the new seeds, re-execute the step of "respectively calculating the distance between each solution in the empirical knowledge base except each of the seeds and each of the seeds" and subsequent steps until all solutions in the empirical knowledge base are traversed to obtain multiple seeds in the empirical knowledge base, wherein each seed represents a mode.

7. The electromechanical system simulation parameter single-condition calibration method according to claim 6, characterized in that: The process of determining the distance threshold is as follows: Calculate the distance between the optimal solution in each subspace and other solutions in the subspace, recorded as the first distance; Determine a representative distance corresponding to each subspace according to each of the first distances in each subspace; The set distance threshold is determined according to the representative distance corresponding to each subspace.

8. A single-condition calibration device for electromechanical system simulation parameters, characterized in that: include: An objective function construction module is used to construct an objective function with the goal of achieving the highest consistency between the simulation data output by the electromechanical system simulation model under the target working condition and the measured data output by the actual electromechanical system; A first optimization module is used to use each simulation parameter of the electromechanical system simulation model as a decision vector, perform single-objective optimization on the objective function, and add the single-objective optimization result to the experience knowledge base; A second optimization module is used to divide each solution of each simulation parameter of the electromechanical system simulation model in the empirical knowledge base into multiple subspaces, and construct a proxy model in each subspace to perform initial Bayesian optimization; A third optimization module is used to add the initial Bayesian optimization result to the empirical knowledge base, perform multimodal detection on each solution in the empirical knowledge base, and construct a proxy model in each mode obtained by the detection to perform secondary Bayesian optimization; A cyclic calibration module is used to add the quadratic Bayesian optimization result to the empirical knowledge base, and re-execute the step of "performing single objective optimization on the objective function" and subsequent steps based on the quadratic Bayesian optimization result until the termination condition is met, and obtain the single working condition calibration result of the simulation parameters of the electromechanical system simulation model according to the single objective optimization result.

9. A terminal, characterized in that: The method comprises a memory and a processor, wherein the memory is used to store a computer program, and the processor is used to call and run the computer program stored in the memory to execute the method according to any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method as claimed in any one of claims 1 to 7 are implemented.

Images