ABS virtual calibration system and method based on hybrid intelligent algorithm

Abstract

The invention relates to the field of ABS virtual calibration, and provides an ABS virtual calibration system and method based on a hybrid intelligent algorithm, and the virtual calibration system is combined with various software to establish a virtual calibration platform which is composed of a to-be-calibrated algorithm module, a parameter calibration module, a virtual vehicle model module, an integrated simulation module and a shared memory module. A virtual vehicle model and a simulation environment are constructed, and dynamic behaviors of the vehicle under different working conditions can be simulated without depending on physical tests, so that adjustment and optimization of system parameters are realized. According to the virtual calibration method, a hybrid intelligent algorithm is designed as an ABS calibration algorithm, a designed ABS evaluation index is used as a design basis of a fitness function, a genetic algorithm and a simulated annealing algorithm are combined, automatic calibration work is carried out on a to-be-calibrated algorithm in an ABS calibration platform, the virtual calibration optimization efficiency is remarkably improved, and the method is suitable for industrial production. And the fidelity is high, the real-time performance is strong, and the technical advantages of virtual calibration can be fully exerted.

Description

Technical Field

[0001] The present invention relates to the field of ABS virtual calibration, and in particular to an ABS virtual calibration system and method based on a hybrid intelligent algorithm. Background Art

[0002] As a core technology for active safety control, the performance of the automotive anti-lock braking system (ABS) directly impacts the vehicle's braking and handling stability. Traditional ABS parameter calibration relies heavily on real-vehicle testing, requiring extensive repeated testing at specialized proving grounds to determine the optimal parameter combination. This approach suffers from three major technical bottlenecks: first, high test site rental costs, prototype vehicle wear and tear, and labor costs, coupled with lengthy calibration cycles; second, the ability to replicate complex operating conditions is constrained by environmental conditions; and third, the nonlinear response characteristics of multiple coupled parameters make it difficult to achieve a globally optimal solution through empirical trial-and-error methods.

[0003] With the development of digital twin technology, virtual calibration technology has been gradually applied to the field of electronic control system development. Existing technical solutions mostly use a single optimization algorithm (such as genetic algorithm, particle swarm algorithm) combined with a vehicle dynamics model. However, the following technical defects have been found in actual applications: (1) The traditional optimization algorithm has a prominent premature convergence phenomenon and is prone to falling into local optimality in the ABS high-dimensional parameter space search; (2) There are data interaction barriers in multi-software collaborative simulation, especially the lack of real-time data channels between embedded control programs and dynamic models, which reduces the simulation iteration efficiency by more than 40%; (3) The weight distribution of evaluation indicators mostly adopts expert experience method, which is highly subjective.

[0004] Due to existing technical limitations, the optimization efficiency of current virtual calibration platforms is limited compared to physical testing, failing to fully leverage the technical advantages of virtual calibration. Therefore, it is urgent to build a new hybrid intelligent algorithm framework for high-fidelity, real-time virtual calibration platforms to overcome the current technical bottlenecks. Summary of the Invention

[0005] In order to solve the above technical problems, the present invention provides an ABS virtual calibration system based on a hybrid intelligent algorithm, which is established through multiple software tools, including a calibration algorithm module, a parameter calibration module, a virtual vehicle model module, an integrated simulation module, and a shared memory module.

[0006] The function of the algorithm module to be calibrated is to compile and run the ABS control program to be calibrated;

[0007] The parameter calibration module includes a parameter adjustment module and a data acquisition module. The parameter adjustment module is used to modify the control parameters in the ABS control program to be calibrated, and the data acquisition module is used to collect key performance indicator data of the integrated simulation module in real time during operation.

[0008] The virtual vehicle model module provides an accurate vehicle dynamics model for ABS simulation, and simulates various typical operating conditions through calibration conditions to support simulation testing during the calibration process. It simulates the dynamic behavior of the vehicle under various operating conditions and simulates the driver's braking force based on the driver model to trigger the ABS working response.

[0009] The integrated simulation module integrates the various modules of the virtual calibration system and realizes linkage operation to simulate the ABS control program to be calibrated;

[0010] The shared memory module is respectively interconnected with the algorithm module to be calibrated and the integrated simulation module to realize data interaction.

[0011] The present invention also provides an ABS virtual calibration method based on a hybrid intelligent algorithm, comprising the following steps:

[0012] Step 1: Design of evaluation indicators and calibration conditions:

[0013] The calibration conditions are common scenarios used in ABS calibration. The calibration target is the weighted minimum value of the braking distance and the braking deviation distance. The weights of each evaluation indicator are calculated based on the entropy weight method.

[0014] As a preference, the steps for calculating the weights of each evaluation index according to the entropy weight method are as follows:

[0015] (1) Raw data collection:

[0016] Collect raw data of performance evaluation indicators from simulation model operation results;

[0017] (2) Normalization processing:

[0018] The original data is normalized using the following formula:

[0019]

[0020] Among them, r ij is the normalized value of the jth index under the i-th working condition, x ij is the original data, max(x j ) and min(x j ) are the maximum and minimum values of the j-th index respectively;

[0021] (2) Calculate information entropy:

[0022] According to the normalized data, the information entropy of each indicator is calculated:

[0023]

[0024] Among them, e jis the information entropy of the jth index, k is a constant, k = 1 / lnn, n is the number of working conditions; when r ij =0, it is specified that r ij ln(r ij )=0;

[0025] (3) Calculate weight:

[0026] The smaller the information entropy, the greater the indicator fluctuation, and the higher the weight should be; the weight calculation formula is as follows:

[0027]

[0028] Among them, w j is the weight of the jth indicator, and m is the total number of indicators.

[0029] Step 2: Basic physical parameter consistency calibration:

[0030] Calibrate the basic physical parameters of the vehicle model to keep them consistent with the corresponding parameters in the ABS control program to be calibrated.

[0031] Step 3: Hybrid intelligent calibration:

[0032] This paper selects a strategy that combines genetic algorithm and simulated annealing algorithm. The genetic algorithm is responsible for extensively exploring the solution space, while the simulated annealing algorithm fine-tunes the candidate solutions, thereby significantly improving the optimization accuracy and convergence speed. The specific steps of hybrid intelligent calibration are as follows:

[0033] (1) Coding and population initialization:

[0034] The encoding method uses real number encoding. After completing chromosome encoding, an initial population of N individuals is randomly generated, where each individual represents a possible parameter combination. The purpose of population initialization is to cover a certain range of solution space through random distribution, providing diversity and good initial conditions for genetic operations. The chromosome of each individual records the parameter value of the ABS control program to be labeled, and its quality is evaluated using the fitness function.

[0035] (2) Fitness function design:

[0036] The evaluation index obtained by entropy weight method is converted into fitness function to quantitatively evaluate the individual;

[0037] As a preference, the fitness function is expressed as:

[0038]

[0039] Among them, F(x) is the fitness value; f i (x) is the value of the i-th performance indicator after normalization; w iis the weight of the determined performance index of item i; m is the total number of performance indexes;

[0040] The fitness function unifies multiple performance indicators into a single optimization objective through comprehensive weighting, enabling the hybrid intelligent calibration algorithm to comprehensively evaluate individuals based on the overall requirements of ABS performance;

[0041] (3) Elite retention strategy:

[0042] In the evolution process of each generation, the elite retention strategy directly retains the individuals with the best fitness in the current population to the next generation without participating in crossover and mutation operations;

[0043] (4) Global search of genetic algorithm:

[0044] The global search of the genetic algorithm is divided into three parts: selection, crossover, and mutation;

[0045] Selection operation: Roulette wheel selection method is used according to the fitness value of the individual, and the selected individual is used as the parent;

[0046] Crossover operation: according to the set crossover probability p c , perform genetic recombination on the selected parent individuals to generate new offspring individuals;

[0047] Mutation operation: with mutation probability p m Randomly change some gene values of offspring individuals to introduce new genetic information;

[0048] Through the synergistic effect of selection, crossover and mutation operations, genetic algorithms can continuously optimize the quality of the population during the global search process, laying the foundation for ultimately obtaining the optimal solution to the problem.

[0049] (5) Local optimization of simulated annealing algorithm:

[0050] Based on the new solutions generated by the genetic algorithm, the individuals with the highest fitness are selected and locally optimized using a simulated annealing algorithm. This algorithm, by introducing a temperature parameter and an acceptance probability mechanism, can effectively escape from local optimal solutions and achieve more refined optimization in the later stages of the search. This is specifically divided into three parts: initialization temperature and cooling strategy, perturbation generation, and acceptance criteria.

[0051] The initialization temperature and cooling strategy set the initial temperature T0, and use the exponential cooling method to perform temperature decay. The temperature update formula is:

[0052] T k+1 =αT k (5)

[0053] Among them, T k+1 is the system temperature at the k+1th iteration; α is the cooling coefficient; Tk is the system temperature at the kth iteration;

[0054] The perturbation generation is to randomly perturb the current solution to explore the local area of the solution space and generate new candidate solutions;

[0055] The acceptance criterion uses the Metropolis criterion to determine whether to replace the current individual with a new individual. The formula is:

[0056]

[0057] Where P is the probability that the current individual is replaced by a new individual; ΔE is the difference between the objective function of the new solution and the current solution; T is the current temperature;

[0058] Closed-loop automatic calibration is performed by calling the corresponding interfaces of the ABS control program to be calibrated and the virtual vehicle model.

[0059] Step 4: Setting the termination conditions:

[0060] The termination conditions of the hybrid intelligent calibration algorithm include the maximum number of iterations or the population fitness change rate meeting the set threshold; when any of the termination conditions is met, the algorithm stops running and outputs the individual with the highest fitness in the current population as the optimization result; otherwise, the algorithm returns to the global search step of the genetic algorithm, enters the evolution of the next generation population, and continues the optimization.

[0061] After the whole process is completed, the corresponding parameter group with the highest fitness function value is selected as the final result of ABS virtual calibration.

[0062] Optionally, the ABS virtual calibration method based on the hybrid intelligent algorithm is executed using the ABS virtual calibration system based on the hybrid intelligent algorithm provided by the present invention.

[0063] Beneficial effects of the present invention:

[0064] The present invention provides an ABS virtual calibration system and method based on a hybrid intelligent algorithm. The virtual calibration system combines multiple software programs to establish a virtual calibration platform consisting of a to-be-calibrated algorithm module, a parameter calibration module, a virtual vehicle model module, an integrated simulation module, and a shared memory module. This system constructs a virtual vehicle model and simulation environment, simulating the dynamic behavior of a vehicle under different operating conditions without relying on physical testing, thereby enabling the adjustment and optimization of system parameters. The virtual calibration method of the present invention designs a hybrid intelligent algorithm as the ABS calibration algorithm, uses the designed ABS evaluation index as the design basis for the fitness function, and combines a genetic algorithm with a simulated annealing algorithm to automatically calibrate the to-be-calibrated algorithm in the ABS calibration platform. This significantly improves virtual calibration optimization efficiency, achieves high fidelity, and enhances real-time performance, fully leveraging the technical advantages of virtual calibration. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 Schematic diagram of the overall architecture of the ABS virtual calibration system based on the hybrid intelligent algorithm of the present invention;

[0066] Figure 2 This is a schematic diagram of a shared memory module of the present invention;

[0067] Figure 3 Schematic diagram of the process of the ABS virtual calibration method based on the hybrid intelligent algorithm of the present invention;

[0068] Figure 4 Schematic diagram of the hybrid intelligent calibration algorithm architecture of the present invention. DETAILED DESCRIPTION

[0069] The present invention provides an ABS virtual calibration system based on a hybrid intelligent algorithm, which is established through a variety of software tools, including a calibration algorithm module, a parameter calibration module, a virtual vehicle model module, an integrated simulation module, and a shared memory module. The specific architecture is as follows: Figure 1 shown.

[0070] The function of the algorithm module to be calibrated is to compile and run the ABS control program to be calibrated. Since most ABS control programs to be calibrated are embedded programs written in C language, the algorithm module to be calibrated is compiled and debugged using Visual Studio.

[0071] The parameter calibration module includes a parameter adjustment module and a data acquisition module. The parameter adjustment module modifies the control parameters in the ABS control program to be calibrated, and the data acquisition module collects key performance indicator data of the integrated simulation module in real time during operation. This module is implemented based on the Python tool chain and completes parameter adjustment and data acquisition tasks by writing Python scripts.

[0072] The virtual vehicle model module provides an accurate vehicle dynamics model for ABS simulation, and simulates various typical operating conditions through calibration conditions to support simulation testing during the calibration process. Based on CarSim, this module accurately simulates the dynamic behavior of the vehicle under various operating conditions. At the same time, it simulates the driver's braking force based on the driver model, thereby triggering the ABS's working response.

[0073] The integrated simulation module integrates the various modules of the virtual calibration system based on Simulink and realizes linkage operation to simulate the ABS control program to be calibrated; Simulink in the integrated simulation module and CarSim in the virtual vehicle model module have their own joint interface, which can realize efficient data interaction.

[0074] However, for the ABS program compiled and run in Visual Studio, due to the lack of a direct interface to interact with Simulink, the present invention adopts shared memory technology, which is an efficient inter-process communication technology. By creating a shared area in the memory, the ABS program in Visual Studio and the Simulink model can exchange data in real time in the area, thereby achieving efficient collaboration between the control algorithm, vehicle dynamics model and parameter calibration module. Figure 2 shown.

[0075] like Figure 3 As shown, the present invention provides an ABS virtual calibration method based on a hybrid intelligent algorithm, comprising the following steps:

[0076] Step 1: Design of evaluation indicators and calibration conditions:

[0077] The calibration condition of this embodiment uses the scenario of emergency braking at 60 km / h on a low-adhesion uniform road surface, which is the most common scenario in ABS calibration. The calibration target is the weighted minimum value of the braking distance and the braking deviation distance. The weight of each evaluation index is calculated according to the entropy weight method. The specific steps are as follows:

[0078] (1) Raw data collection:

[0079] The parameter calibration module collects the original data of performance evaluation indicators from the simulation model running results of the integrated simulation module;

[0080] (2) Normalization processing:

[0081] The original data is dimensionless processed to eliminate the influence caused by the difference in dimensions and magnitudes between different indicators. The normalization formula is as follows:

[0082]

[0083] Among them, r ij is the normalized value of the jth index under the i-th working condition, x ij is the original data, max(x j ) and min(x j ) are the maximum and minimum values of the j-th index respectively;

[0084] (2) Calculate information entropy:

[0085] According to the normalized data, the information entropy of each indicator is calculated:

[0086]

[0087] Among them, e jis the information entropy of the jth index, k is a constant, k = 1 / lnn, n is the number of working conditions; when r ij =0, it is specified that r ij ln(r ij )=0;

[0088] (3) Calculate weight:

[0089] The smaller the information entropy, the greater the indicator fluctuation, and the higher the weight should be; the weight calculation formula is as follows:

[0090]

[0091] Among them, w j is the weight of the jth indicator, and m is the total number of indicators.

[0092] Step 2: Basic physical parameter consistency calibration:

[0093] To ensure that the simulation results truly reflect the behavior of the actual vehicle, the basic physical parameters of the vehicle model in the virtual vehicle model module (CarSim) must be strictly calibrated to ensure that they are consistent with the corresponding parameters in the ABS control program to be calibrated. The calibration parameter table in this embodiment is the ABS built-in parameter table to be calibrated, see Table 1.

[0094] Step 3: Hybrid intelligent calibration:

[0095] Hybrid intelligent calibration algorithm architecture Figure 4 As shown, the present invention selects a strategy that combines genetic algorithm and simulated annealing algorithm, in which the genetic algorithm is responsible for extensively exploring the solution space, while the simulated annealing algorithm fine-tunes the candidate solutions, thereby significantly improving the optimization accuracy and convergence speed. The specific steps of hybrid intelligent calibration are as follows:

[0096] (1) Coding and population initialization:

[0097] Encoding and population initialization operations are required before hybrid intelligent calibration is implemented.

[0098] The encoding method uses real number encoding. Compared with binary encoding, real number encoding has higher accuracy and can directly represent continuous parameters, avoiding the accuracy loss caused by binary decoding.

[0099] After completing chromosome encoding, an initial population of N individuals is randomly generated, where each individual represents a possible parameter combination. Population initialization aims to cover a wider solution space through random distribution, providing diversity and good initial conditions for genetic operations. The chromosome of each individual records the parameter values of the ABS control program to be labeled, and its quality is evaluated through the fitness function, laying the foundation for the subsequent evolutionary process.

[0100] (2) Fitness function design:

[0101] The evaluation index obtained after entropy weighting is converted into a fitness function to quantitatively evaluate the individual. The expression of the fitness function is:

[0102]

[0103] Among them, F(x) is the fitness value; f i (x) is the value of the i-th performance indicator after normalization; w i is the weight of the determined i-th performance indicator; m is the total number of performance indicators;

[0104] The fitness function unifies multiple performance indicators into a single optimization objective through comprehensive weights, enabling the hybrid intelligent calibration algorithm to comprehensively evaluate individuals based on the overall requirements of ABS performance. The larger the fitness value, the better the overall performance of the individual, and the more likely it is to be retained in the selection operation or used to generate offspring.

[0105] (3) Elite retention strategy:

[0106] During the evolution of each generation, the elite retention strategy directly retains the individuals with the best fitness in the current population to the next generation without participating in crossover and mutation operations.

[0107] (4) Global search of genetic algorithm:

[0108] The global search of the genetic algorithm is divided into three parts: selection, crossover, and mutation;

[0109] Selection: Roulette wheel selection is used based on the fitness of the individual, and the selected individuals serve as parents. Individuals with high fitness are more likely to be selected, thus ensuring that good genes are passed on and enhancing the overall adaptability of the population.

[0110] Crossover operation: according to the set crossover probability p c , genetically recombining the selected parent individuals to generate new offspring individuals. The crossover operation explores a broader solution space by fusing the genetic information of the parent generation, increasing the potential diversity of the population and providing more optimization directions for the algorithm.

[0111] Mutation operation: with mutation probability p m Randomly changing the values of certain genes in offspring individuals introduces new genetic information. Mutation can effectively prevent the population from falling into a local optimal solution, enhance the diversity of the population, and improve the algorithm's ability to explore the global solution space.

[0112] Through the synergistic effect of selection, crossover and mutation operations, genetic algorithms can continuously optimize the quality of the population during the global search process, laying the foundation for ultimately obtaining the optimal solution to the problem.

[0113] (5) Local optimization of simulated annealing algorithm:

[0114] Based on the new solutions generated by the genetic algorithm, the individuals with the highest fitness are selected and locally optimized using the simulated annealing algorithm. By introducing a temperature parameter and an acceptance probability mechanism, the simulated annealing algorithm can effectively escape local optimal solutions and achieve more refined optimization in the later stages of the search. This algorithm is specifically divided into three parts: initialization temperature and cooling strategy, perturbation generation, and acceptance criteria.

[0115] The initialization temperature and cooling strategy set the initial temperature T0, and use the exponential cooling method to perform temperature decay. The temperature update formula is:

[0116] T k+1 =αT k (5)

[0117] Among them, T k+1 is the system temperature at the k+1th iteration; α is the cooling coefficient; T k is the system temperature at the kth iteration. The initial temperature setting must ensure that the algorithm has a large search range in the initial stage, and the cooling strategy determines the convergence speed and accuracy of the algorithm.

[0118] The perturbation generation is to perform random perturbations on the current solution to explore local areas of the solution space and generate new candidate solutions.

[0119] The acceptance criterion is to avoid the algorithm from falling into a local optimal solution. The Metropolis criterion is applied to determine whether to replace the current individual with a new individual. The formula is:

[0120]

[0121] Where P is the probability that the current individual is replaced by a new individual; ΔE is the difference in the objective function between the new solution and the current solution; and T is the current temperature. The simulated annealing algorithm allows for suboptimal solutions by accepting probabilities, effectively avoiding local optima and achieving more refined optimization in the later stages of the search.

[0122] After the hybrid intelligent calibration algorithm is implemented, the integrated simulation module in the virtual calibration system designed by the present invention can call the corresponding interfaces of the ABS control program to be calibrated in the algorithm module to be calibrated and the virtual vehicle model module to perform closed-loop automatic calibration.

[0123] Step 4: Setting the termination conditions:

[0124] The termination conditions of the hybrid intelligent calibration algorithm include the maximum number of iterations or the population fitness change rate meeting the set threshold; when any of the termination conditions is met, the algorithm stops running and outputs the individual with the highest fitness in the current population as the optimization result; otherwise, the algorithm returns to the global search step of the genetic algorithm, enters the evolution of the next generation population, and continues the optimization.

[0125] After the whole process is completed, the corresponding parameter group with the highest fitness function value is selected as the final result of ABS virtual calibration.

Claims

1. An ABS virtual calibration method based on a hybrid intelligent algorithm, characterized by: The following steps are involved: Step 1: Design of evaluation indicators and calibration conditions: The calibration conditions are common scenarios used in ABS calibration. The calibration target is the weighted minimum value of the braking distance and the braking deviation distance. The weights of each evaluation indicator are calculated based on the entropy weight method. Step 2: Basic physical parameter consistency calibration: Calibrate the basic physical parameters of the vehicle model to make them consistent with the corresponding parameters in the ABS control program to be calibrated; Step 3: Hybrid intelligent calibration: The genetic algorithm is combined with the simulated annealing algorithm, wherein the genetic algorithm is responsible for extensively exploring the solution space, and the simulated annealing algorithm is responsible for fine-tuning the candidate solutions. The steps of hybrid intelligent calibration are as follows: (1) Coding and population initialization: The encoding method uses real number encoding. After completing chromosome encoding, an initial population consisting of N individuals is randomly generated, where each individual represents a parameter combination. The population is initialized by randomly distributing the solution space to cover a certain range, providing initial conditions for genetic operations. The chromosome of each individual records the parameter value of the ABS control program to be labeled, and its quality is evaluated using the fitness function. (2) Fitness function design: The evaluation index obtained by entropy weight method is converted into fitness function to quantitatively evaluate the individual; (3) Elite retention strategy: In the evolution process of each generation, the elite retention strategy directly retains the individuals with the best fitness in the current population to the next generation without participating in crossover and mutation operations; (4) Global search of genetic algorithm: The global search of the genetic algorithm is divided into three parts: selection, crossover, and mutation; Selection operation: Roulette wheel selection method is used according to the fitness value of the individual, and the selected individual is used as the parent; Crossover operation: according to the set crossover probability p c , perform genetic recombination on the selected parent individuals to generate new offspring individuals; Mutation operation: with mutation probability p m Randomly change some gene values of offspring individuals to introduce new genetic information; (5) Local optimization of simulated annealing algorithm: Based on the new solution generated by the genetic algorithm, the individual with the highest fitness is selected and locally optimized using a simulated annealing algorithm. The simulated annealing algorithm introduces a temperature parameter and an acceptance probability mechanism, and is divided into three parts: initialization temperature and cooling strategy, perturbation generation, and acceptance criteria. The initialization temperature and cooling strategy set the initial temperature T0, and use the exponential cooling method to perform temperature decay. The temperature update formula is: T k+1 =αT k Among them, T k+1 is the system temperature at the k+1th iteration; α is the cooling coefficient; T k is the system temperature at the kth iteration; The perturbation generation is to randomly perturb the current solution to explore the local area of the solution space and generate new candidate solutions; The acceptance criterion uses the Metropolis criterion to determine whether to replace the current individual with a new individual. The formula is: Where P is the probability that the current individual is replaced by a new individual; ΔE is the difference between the objective function of the new solution and the current solution; T is the current temperature; Perform closed-loop automatic calibration by calling the ABS control program to be calibrated and the virtual vehicle model; Step 4: Setting the termination conditions: The termination conditions of the hybrid intelligent calibration algorithm include the maximum number of iterations or the population fitness change rate meeting the set threshold. When any of the termination conditions is met, the algorithm stops running and outputs the individual with the highest fitness in the current population as the optimization result. Otherwise, the algorithm returns to the global search step of the genetic algorithm and enters the next generation of population evolution to continue optimization. After the whole process is completed, the corresponding parameter group with the highest fitness function value is selected as the final result of ABS virtual calibration.

2. The ABS virtual calibration method based on a hybrid intelligent algorithm according to claim 1, characterized in that: In the first step, the steps for calculating the weights of each evaluation index based on the entropy weight method are as follows: (1) Raw data collection: Collect raw data of performance evaluation indicators from simulation model operation results; (2) Normalization processing: The original data is normalized using the following formula: Among them, r ij is the normalized value of the jth index under the i-th working condition, x ij is the original data, max(x j ) and min(x j ) are the maximum and minimum values of the j-th index respectively; (2) Calculate information entropy: According to the normalized data, the information entropy of each indicator is calculated: Among them, e j is the information entropy of the jth index, k is a constant, k = 1 / lnn, n is the number of working conditions; when r ij =0, it is specified that r ij ln(r ij )=0; (3) Calculate weight: The smaller the information entropy, the greater the indicator fluctuation, and the higher the weight should be; the weight calculation formula is as follows: Among them, w j is the weight of the jth indicator, and m is the total number of indicators.

3. The ABS virtual calibration method based on a hybrid intelligent algorithm according to claim 1, characterized in that: The expression of the fitness function described in the third step is: Among them, F(x) is the fitness value; f i (x) is the value of the i-th performance indicator after normalization; w i is the weight of the determined performance evaluation item i; m is the total number of performance indicators.

4. An ABS virtual calibration system based on a hybrid intelligent algorithm, characterized by: It includes algorithm module to be calibrated, parameter calibration module, virtual vehicle model module, integrated simulation module and shared memory module; The function of the algorithm module to be calibrated is to compile and run the ABS control program to be calibrated; The parameter calibration module includes a parameter adjustment module and a data acquisition module. The parameter adjustment module is used to modify the control parameters in the ABS control program to be calibrated, and the data acquisition module is used to collect key performance indicator data of the integrated simulation module in real time during operation. The virtual vehicle model module provides an accurate vehicle dynamics model for ABS simulation, and simulates various typical operating conditions through calibration conditions to support simulation testing during the calibration process. It simulates the dynamic behavior of the vehicle under various operating conditions and simulates the driver's braking force based on the driver model to trigger the ABS working response. The integrated simulation module integrates the various modules of the virtual calibration system and realizes linkage operation to simulate the ABS control program to be calibrated; The shared memory module is interconnected with the algorithm module to be calibrated and the integrated simulation module to realize data interaction; The ABS virtual calibration system based on the hybrid intelligent algorithm executes the ABS virtual calibration method based on the hybrid intelligent algorithm described in any one of claims 1 to 3.

Images