A tire model parameter identification method, device, equipment and storage medium

Abstract

Embodiments of the present application provide a tire model parameter identification method, device and equipment and a storage medium, wherein the tire model parameter identification method comprises: obtaining a tire model parameter to be identified of a tire model; generating an initialized population according to the tire model parameter to be identified; generating a historical successful experience archive; the historical successful experience archive is used for controlling the evolution direction of the initialized population; controlling the initialized population to perform iterative evolution according to the historical successful experience archive; and determining a target tire model parameter corresponding to the tire model according to the result of the iterative evolution. The method can accurately identify the tire model parameter corresponding to the tire model, and adaptively adjusts according to the current situation in the identification process, thereby improving the accuracy, generalization ability and robustness of the parameter identification result.

Description

Technical Field

[0001] This application relates to the field of vehicle technology, and in particular to a tire model parameter identification method and apparatus, an electronic device, and a storage medium. Background Technology

[0002] Establishing a mathematical model that accurately describes tire-road interaction is a crucial foundation for vehicle dynamics simulation, active safety control, and the development of autonomous driving systems. Among numerous tire models, the Magic Formula (MF) model has become the mainstream choice in both industry and academia due to its versatility and high accuracy. The accuracy of the Magic Formula model highly depends on the precise identification of its parameters. During parameter identification, due to its highly nonlinear and multimodal characteristics, traditional optimization methods often struggle to effectively converge to the global optimum.

[0003] In related technologies, to improve the accuracy of tire model magic formula parameter identification, methods have been proposed, including a two-stage parameter identification process, using nonlinear least squares to find the optimal estimate of initial values, followed by further optimization using particle swarm optimization for first-level parameter identification, and using a genetic algorithm to directly identify second-level parameters. However, these technologies still struggle to adaptively adjust key parameters under complex conditions, accurately describe the dynamic characteristics of the tire, and fail to balance global diversity with local convergence, resulting in a lack of accuracy, generalization ability, and robustness in the final identification results.

[0004] Therefore, how to enable the parameter identification of the tire model magic formula to be adaptively adjusted according to specific circumstances, so as to improve the accuracy, generalization ability and robustness of the parameter identification results, has become an urgent problem to be solved. Summary of the Invention

[0005] This application provides a tire model parameter identification method to solve the problems of difficulty in accurately identifying tire model parameters under complex conditions and the lack of generalization ability and robustness in the identification results.

[0006] Accordingly, embodiments of this application also provide a tire model parameter identification device, an electronic device, and a storage medium to ensure the implementation and application of the above method.

[0007] To address the aforementioned problems, this application discloses a tire model parameter identification method, the method comprising:

[0008] Obtain the parameters of the tire model to be identified;

[0009] An initial population is generated based on the parameters of the tire model to be identified;

[0010] Generate a historical success experience archive; the historical success experience archive is used to control the evolutionary direction of the initialized population;

[0011] Based on the archived historical successes, the initialized population is controlled to undergo iterative evolution.

[0012] The target tire model parameters corresponding to the tire model are determined based on the results of the iterative evolution.

[0013] Optionally, generating an initialized population based on the parameters of the tire model to be identified includes:

[0014] Based on the parameters of the tire model to be identified, a preset number of uniformly distributed random values ​​are selected from the preset search space as the population individuals; the dimension of the population individuals corresponds to the number of parameters of the tire model to be identified; the initialized population includes a preset number of population individuals.

[0015] Optionally, the population consists of multiple individuals, and the step of controlling the initial population to undergo iterative evolution based on the archived historical success experiences includes the following steps:

[0016] Generate the current fitness value evaluation count and the maximum fitness value evaluation count;

[0017] The initialized population is defined as the current population;

[0018] Based on the archived historical success experiences, the control parameters corresponding to the individuals in the current population are determined.

[0019] The mutation vectors corresponding to the individuals in the current population are determined based on the control parameters.

[0020] The crossover vectors corresponding to the individuals in the contemporary population are determined based on the mutation vectors.

[0021] The fitness value of the crossover vector is evaluated to obtain the preferred vector of the current population;

[0022] Update the historical success experience archive and the current population according to the preferred vector;

[0023] Update the current fitness value evaluation count;

[0024] If the number of times the current fitness value is evaluated is less than the maximum number of times the fitness value is evaluated, return to the control parameters corresponding to the population individuals in the current population determined based on the archive of historical successful experiences;

[0025] If the current fitness value evaluation count equals the maximum fitness value evaluation count, output the result of the iterative evolution.

[0026] Optionally, the historical success experience archive includes the average crossover rate and average mutation factor corresponding to the individuals of the population that have successfully evolved, and the average crossover rate and average mutation factor preset when the historical success experience archive was generated. The step of determining the control parameters corresponding to the individuals of the current population based on the historical success experience archive includes:

[0027] The average crossover rate is randomly sampled to obtain the sampled crossover rate;

[0028] The adaptive crossover rate for each individual in the current generation of the population is generated based on the sampling crossover rate.

[0029] Random sampling is performed on the average mutation factor to obtain the sampled mutation factor;

[0030] The adaptive mutation factor for each individual in the current population is generated based on the sampled mutation factor; the control parameters include the adaptive crossover rate and the adaptive mutation factor.

[0031] Optionally, determining the mutation vector corresponding to the individual in the current population based on the control parameters includes:

[0032] The mutation operator to be used is determined based on the number of evaluations of the current fitness value at the start of the evolution of the contemporary population, and the mutation operator includes global mutation operators and local mutation operators;

[0033] Based on the control parameters and the mutation operator used, the individuals in the population are mutated to obtain the mutation vector.

[0034] Optionally, determining the crossover vector corresponding to the individuals in the current population based on the mutation vector includes:

[0035] Based on the mutation vector and the current vector corresponding to the individual in the population at the start of the contemporary population evolution, determine each dimension of the crossover vector;

[0036] The cross vector is generated based on each dimension of the cross vector.

[0037] Optionally, updating the historical success experience archive and the current population according to the preferred vector includes:

[0038] The preferred vectors whose fitness values ​​are less than a preset threshold are identified as population individuals that have successfully evolved in the contemporary population.

[0039] Calculate the average crossover rate and average mutation factor for the individuals in the successfully evolved population.

[0040] The historical success experience archive is updated based on the average crossover rate and the average mutation factor.

[0041] The next generation of individuals and individuals to be eliminated are determined based on the fitness values ​​corresponding to the preferred vectors.

[0042] From the individuals in the population to be eliminated, select individuals with potential potential populations;

[0043] The current population is updated based on the potential population individuals and the next generation population individuals, wherein the potential population individuals are population individuals whose current fitness value is less than that of the previous generation and are to be eliminated.

[0044] This application also discloses a tire model parameter identification device, the device comprising:

[0045] The acquisition module is used to acquire the parameters of the tire model to be identified.

[0046] The population generation module is used to generate an initial population based on the parameters of the tire model to be identified.

[0047] A historical success experience archive generation module is used to generate a historical success experience archive; the historical success experience archive is used to control the evolutionary direction of the initialized population;

[0048] The iterative evolution module is used to control the initial population to undergo iterative evolution based on the archived historical success experiences.

[0049] The determination module is used to determine the target tire model parameters corresponding to the tire model based on the results of the iterative evolution.

[0050] This application also discloses an electronic device, including: a processor; and a memory storing executable code thereon, which, when executed, causes the processor to perform one or more of the tire model parameter identification methods described in this application.

[0051] This application also discloses one or more machine-readable media storing executable code thereon, which, when executed, causes a processor to perform one or more tire model parameter identification methods as described in this application.

[0052] Compared with the prior art, the embodiments of this application have the following advantages:

[0053] In this embodiment, the parameters of the tire model to be identified are obtained. An initial population is generated based on these parameters. A historical success experience archive is generated to control the evolutionary direction of the initial population. Then, the initial population can be iteratively evolved based on the historical success experience archive. The target tire model parameters corresponding to the tire model are determined based on the results of the iterative evolution, achieving high-precision identification of the tire model parameters corresponding to the tire model. The generated historical success experience archive allows for adaptive adjustment during the evolutionary process, resulting in better generalization ability and robustness of the final identification result, thereby improving the accuracy of the corresponding tire model. Attached Figure Description

[0054] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0055] Figure 1 This is a flowchart illustrating the steps of an embodiment of a tire model parameter identification method according to this application;

[0056] Figure 2 This is a side slip angle-lateral force curve after unit conversion, representing an embodiment of the tire model parameter identification method of this application.

[0057] Figure 3 This is a side slip angle-lateral force curve diagram with unit conversion and expansion, representing an embodiment of a tire model parameter identification method of this application.

[0058] Figure 4 This is a flowchart of the mutation strategy scheduling mechanism of an embodiment of the tire model parameter identification method of this application;

[0059] Figure 5 This is a flowchart illustrating the adaptive population size reduction of an embodiment of the tire model parameter identification method of this application;

[0060] Figure 6 This is an identification effect diagram of an embodiment of the tire model parameter identification method of this application;

[0061] Figure 7 This is a flowchart illustrating the steps of another embodiment of the tire model parameter identification method of this application;

[0062] Figure 8 This is a structural block diagram of an embodiment of a tire model parameter identification device according to this application. 。 Detailed Implementation

[0063] To make the above-mentioned objectives, features, and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0064] As the only contact component between a vehicle and the road, the tire is a core element in vehicle dynamics modeling and control. Tire parameters reflect complex physical phenomena such as elastic deformation, slippage, and friction during tire-road contact, and the tire's mechanical properties directly affect vehicle handling stability and ride comfort. However, the tire itself is a highly nonlinear mechanical system, and the forces and moments of its interaction with the road surface, such as lateral forces, longitudinal forces, and self-aligning torques, are difficult to derive directly from theoretical formulas. Therefore, establishing a mathematical model that accurately describes the tire-road interaction is a crucial foundation for vehicle dynamics simulation, active safety control systems such as Anti-lock Braking System (ABS), Electronic Stability Program (ESP), and the development of autonomous driving systems.

[0065] Among numerous tire models, the Magic Formula (MF) model has become the mainstream choice in both industry and academia due to its versatility and high precision. The Magic Formula model is specifically represented as follows:

[0066] Y(x)=Dsin{Carctan[Bx-E(Bx-arctan(Bx))]}

[0067] Y(X)=Y(x)+S v

[0068] x = X + S H

[0069] Where x represents the corrected roll angle or corrected longitudinal slip ratio considering horizontal offset, Y(x) represents the reference lateral force calculated based on the corrected roll angle or the reference longitudinal force calculated based on the corrected longitudinal slip ratio, Y(X) represents the actual lateral force or actual longitudinal force considering vertical and horizontal offset, and X represents the original roll angle or original longitudinal slip ratio. B is the stiffness factor, B = K / (CD). K = tanθ, where K is the tire lateral stiffness or longitudinal slip stiffness, corresponding to the slope at the origin of the curve, and θ is the angle between the tangent of the tire mechanical characteristic curve at the origin and the horizontal axis. C is the shape factor, representing the range of the sine function independent variable in the magic formula, thus determining the shape of the final curve. D is the peak factor, which determines the peak value of the curve. E is the curvature factor, which controls the curvature at the peak value of the curve. S H For horizontal offset, it expresses the amount of horizontal shift of the curve. S V Vertical offset expresses the amount of translation of the curve along the vertical direction.

[0070] The accuracy of the Magic Formula model heavily relies on the precise identification of its parameters. However, due to its highly nonlinear and multimodal characteristics, traditional optimization methods often struggle to converge effectively to the global optimum during parameter identification. Parameter identification in the Magic Formula involves multiple coupled variables and requires accurately capturing complex tire mechanics in a high-dimensional space, resulting in high computational complexity and difficulty. Therefore, selecting an optimization algorithm capable of effectively exploring the complex solution space is crucial. The algorithm must maintain a balance between global search and local convergence, avoiding getting trapped in local optima, and thus efficiently and accurately identifying the optimal parameter combination.

[0071] To improve the accuracy of tire model magic formula parameter identification, existing technologies offer various methods, including: identifying magic formula parameters in a two-stage process; using nonlinear least squares to find the best estimate of initial values ​​and then using particle swarm optimization to further optimize and obtain first-level parameter values; using genetic algorithms to directly identify second-level parameters; and using multi-population genetic algorithms and GUI (Graphical User Interface) to complete the identification of magic formula second-level parameters.

[0072] However, most related technologies focus on parameter identification for older versions of the Magic Formula model, such as Pacejka89, Pacejka94, and PAC2002. Traditional Magic Formula models are primarily used for handling stability analysis under steady-state conditions. Their performance is limited in handling complex transient conditions, high-frequency dynamic responses, and the coupling effects of multiple physical characteristics, making it difficult to accurately describe the dynamic characteristics of tires under complex conditions. The hyperparameters of the optimization algorithms used in these technologies almost entirely rely on empirical settings, lacking the ability to adaptively adjust key parameters according to the specific characteristics of the problem during iteration. This results in limited ability to balance global diversity and local convergence, failing to effectively solve Magic Formula problems with highly nonlinear characteristics and complex solution spaces. When identifying parameters for Magic Formulas, related technologies employ a staged approach to improve solution accuracy, which may lead to overfitting of the identification results, resulting in a lack of necessary generalization ability and robustness. Furthermore, this type of method increases the complexity of the identification process. These factors can all lead to a lack of accuracy, generalization ability, and robustness in the final identification results, thereby affecting the accuracy of the tire model.

[0073] To address the aforementioned issues, this application proposes a tire model parameter identification method that can adaptively adjust parameters during the identification process, resulting in better accuracy, generalization ability, and robustness of the final identification results, thereby improving the precision of the corresponding tire model.

[0074] Reference Figure 1 This is a flowchart illustrating the steps of an embodiment of a tire model parameter identification method according to this application, including the following steps:

[0075] Step 101: Obtain the parameters of the tire model to be identified;

[0076] The tire model can be an MF-Tyre (Magic Formula Tyre) model. The Magic Formula uses a combination of trigonometric functions to fit tire test data. A single set of formulas can completely express the longitudinal force, lateral force, self-aligning torque, rollover torque, drag torque, and the combined effect of longitudinal and lateral forces. The MF-Tyre model is suitable for product design, automotive dynamics simulation, and experimental comparisons where precise descriptions of tire mechanical properties are required. It helps engineers predict vehicle performance under extreme conditions.

[0077] Step 102: Generate an initialized population based on the parameters of the tire model to be identified;

[0078] Step 103: Generate a historical success experience archive; the historical success experience archive is used to control the evolutionary direction of the initialized population;

[0079] Step 104: Based on the archived historical success experiences, control the initialized population to undergo iterative evolution;

[0080] Iterative evolution can correspond to various evolutionary algorithms. For example, the improved L-SHADE (Linear Population Size Adaptation in Differential Evolution with SHADE algorithm) algorithm can be used.

[0081] Step 105: Determine the target tire model parameters corresponding to the tire model based on the results of the iterative evolution.

[0082] In one embodiment provided in this application, tire model parameters to be identified can be obtained, an initial population can be generated based on the tire model parameters, and a historical success experience archive can be generated to control the evolutionary direction of the initial population. Then, the initial population can be controlled to undergo iterative evolution based on the historical success experience archive. The target tire model parameters corresponding to the tire model can be determined based on the results of the iterative evolution, thereby achieving high-precision identification of the tire model parameters corresponding to the tire model. Based on the generated historical success experience archive, adaptive adjustments can be made during the evolutionary process, so that the final identification result has better generalization ability and robustness, thereby improving the accuracy of the corresponding tire model.

[0083] In one embodiment provided in this application, obtaining the tire model parameters to be identified from the tire model includes:

[0084] Acquire the mechanical response data of the tire-ground interaction and its corresponding experimental parameters;

[0085] The mechanical response data of the tire-ground interaction can include lateral force, longitudinal force, and self-aligning torque, and the corresponding experimental parameters can be sideslip angle, longitudinal slip ratio, vertical load, etc.

[0086] Outlier data points are removed from the mechanical response data to obtain valid mechanical response data;

[0087] A sample of effective tire parameters is obtained based on the effective mechanical response data and its corresponding experimental parameters;

[0088] The effective tire parameter sample may include effective mechanical response data and its corresponding experimental parameters and time series.

[0089] The corresponding tire model parameters to be identified are determined based on the valid tire parameter samples.

[0090] When collecting mechanical response data of tire-ground interaction, a data acquisition system based on a tire dynamic characteristic test bench can be used. High-precision sensors can be used to acquire this data in real time. A high-precision six-component force sensor can be equipped on the test bench. The sensor's measurement range covers lateral force, longitudinal force, and self-aligning torque under typical tire operating conditions, with a sampling frequency of no less than 1 kHz and a signal-to-noise ratio better than 60 dB, ensuring high fidelity in the dynamic response. During the test, the tire slip angle can be precisely applied using the test bench's electric cylinder or hydraulic servo system, with a control accuracy better than ±0.1°, and a dynamic response frequency range covering 0-30 Hz. The longitudinal slip ratio can be achieved by adjusting the difference between the test bench's roller speed and the tire speed. The vertical load can be accurately applied using a hydraulic servo system and remains constant throughout the experiment. Based on the input tire slip angle, longitudinal slip ratio, and vertical load, the corresponding sensor output raw signals are obtained as the mechanical response data. After acquiring the raw sensor signal, wavelet transform can be used to process it, and outlier data points can be removed by setting a threshold to denoise the raw sensor signal, retaining only the effective mechanical response data that conforms to the physical characteristics of the tire. The processed data can be stored in a standardized format, such as a CSV (Comma-Separated Values) file or MATLAB's standard data storage format, MAT. The file can contain time series, mechanical response data, effective mechanical response data, and experimental parameters, providing high-quality input for subsequent Magic Formula parameter identification.

[0091] For example, taking the MF-Tyre lateral force model as an example, its formula is as follows:

[0092] F y =D y sin(C y arctan(B y α y -E y (B y α y -arctan(B y α y ))))+S Vy

[0093]

[0094] C y =P Cy1

[0095] μ y =(P Dy1 +P Dy2 dfz (1-P) Dy3 γ 2 )

[0096] D y =μ y F z

[0097]

[0098] K yγ0 =F z (P Ky6 +P Ky7 df z )

[0099] S Vyγ =F z (P Vy3 +P Vy4 df z )γ

[0100] S Vy =F z (P Vy1 +P Vy2 df z )+S Vyγ

[0101]

[0102] α y =α+S Hy

[0103] E y =(P Ey1 +P Ey2 df z ){1+P Ey5 γ 2 -(P Ey4 +P Ey3 γ)sgn(α y )}

[0104] Among them, F y The force is lateral force; α is the tire slip angle; γ is the tire camber angle; F Z For the vertical load on the wheel; F Z0 For reference load; F zn For normalized vertical load; df z μ is the deviation rate of the vertical load relative to the reference value. y K represents the peak adhesion coefficient for lateral force. yα For tire lateral stiffness; K yγ0 S represents tire camber stiffness. Vyγ B is the vertical offset caused by the tire camber angle.y C is the stiffness factor. y D is the shape factor; y E is the peak factor; y S is the curvature factor; Hy S is the horizontal offset. Vy α is the vertical offset. y P is the tire's equivalent slip angle; Cy1 P Dy1 P Dy2 P Dy3 P Ey1 P Ey2 P Ey3 P Ey4 P Ey5 P Ky1 P Ky2 P Ky3 P Ky4 P Ky5 P Ky6 P Ky7 P Vy1 P Vy2 P Vy3 P Vy4 P Hy1 and P Hy2 These are the P-type parameters for the tire model.

[0105] Typically, C y D y E y S Hy S Vy These are called first-order parameters, used to describe a certain tilt angle γ and vertical load F. Z Under the action of the lateral force F y The parameters change with the sideslip angle α. Class P parameters are called secondary parameters, corresponding to the primary parameters as a function of the camber angle γ and vertical load, respectively. The subscript of P represents the corresponding primary parameter. The Class P parameters of the tire model are identified as the tire model parameters to be identified.

[0106] After determining the tire parameters to be identified, the units of the collected data can be converted. For example, the units of slip angle, roll angle, vertical load, and lateral force can be converted to radians (rad), radians (rad), radians (N), and radians (N).

[0107] Reference Figure 2 The diagram shown is a side slip angle-lateral force curve after unit conversion, and Figure 3The diagram shows the slip angle-lateral force curve after unit conversion and expansion. To ensure the tire lateral force model has symmetry and robustness in both the positive and negative slip angle directions, the original experimental data can be negatively expanded. The original experimental data consists of slip angles and lateral forces under different vertical loads and lateral roll angles. With a lateral roll angle of 0, the curves corresponding to different vertical loads in the original experimental data can be plotted as shown below. Figure 2 The side slip angle-lateral force curve shown is after unit conversion. For each side slip angle and its corresponding lateral force, a negative value can be generated, thus constructing a complete side slip angle-lateral force dataset. This processing is based on the assumption of tire mechanics symmetry, meaning that when the side slip angle direction reverses, the lateral force direction also reverses synchronously, and its absolute value remains unchanged. By negatively expanding the original experimental parameters, the sample size of the original dataset can be doubled, while simultaneously covering the entire input range of both positive and negative side slip angles. Based on the expanded side slip angle-lateral force dataset, curves such as... can be plotted. Figure 3 The side slip angle-lateral force curve is shown after unit conversion and expansion. Figure 2 and Figure 3 The curves in the diagram correspond to the 1st to 8th types of vertical loads, with the side slip angle-lateral force curve corresponding to a different color for each type of vertical load, based on the degree of curvature from smallest to largest.

[0108] In one embodiment provided in this application, an objective function can be constructed to determine the population individuals that have successfully evolved. This objective function can be constructed based on a multi-dimensional evaluation using goodness of fit and mean absolute percentage error (MASE). Goodness of fit and MASE are calculated from valid tire parameter samples. Goodness of fit ensures that the overall trend of the model is consistent with the data, while MASE effectively suppresses the excessive amplification of errors by outliers, avoiding the limitations of a single indicator. Based on the above implementation process, combining the global interpretability of goodness of fit with the robust error measure of MASE can overcome the limitations of a single indicator in model evaluation, accurately assess the absolute scale of prediction errors, and thus achieve dual verification of the model's global fitting quality and local prediction accuracy.

[0109] For example, the objective function is set as follows:

[0110] Fitness = w1 * (1-R) 2 )+w2*MAPE

[0111] Where Fitness is the objective function, R 2 For goodness of fit, MAPE is the mean absolute percentage error, and w1 and w2 are weighting coefficients.

[0112] Goodness of fit R 2 The calculation formula is as follows:

[0113]

[0114] The formula for calculating the Mean Absolute Percentage Error (MAPE) is as follows:

[0115]

[0116] Where i = 1, ..., n, n represents the number of samples of vertical load extracted, j = 1, ..., m, m represents the number of samples of tire roll angle extracted, and k = 1, ..., g, g represents the number of samples of tire slip angle extracted. The number of samples can be set according to actual conditions, and this application does not impose specific limitations. F Z F represents the vertical load on the wheel, γ is the tire camber angle, and α is the tire slip angle. Data F represents the actual observed value. Model P represents the model's predicted value. Model This represents the model confidence level.

[0117] In one embodiment of this application, generating an initialized population based on the parameters of the tire model to be identified includes:

[0118] Based on the parameters of the tire model to be identified, a preset number of uniformly distributed random values ​​are selected from the preset search space as the population individuals; the dimension of the population individuals corresponds to the number of parameters of the tire model to be identified; the initialized population includes a preset number of population individuals.

[0119] In this population, each individual corresponds to a solution vector. Based on the number of parameters in the tire model to be identified, a predetermined number of uniformly distributed random values ​​can be selected from the pre-defined search space as individuals in the population. The vector dimension of each individual corresponds to the number of parameters in the tire model to be identified. The initialized population includes a predetermined number of individuals. The number of individuals in the population, i.e., the population size, can typically be set to 5 to 10 times the vector dimension; this application does not impose specific limitations on this.

[0120] For example, the population individuals in the initial population can be generated using the following formula.

[0121]

[0122] Where L is the lower boundary of the predefined search space, and U is the upper boundary of the predefined search space. The corresponding search boundary can be set to [-1000, 1000], which is not specifically limited in this application. rand∈[0,1] is a uniformly distributed random number, i=1,2,……,N, where N is the population size. j=1,2,……,D, where D is the dimension of the solution vector corresponding to the individual in the population, that is, the number of parameters of the tire model to be identified.

[0123] In one embodiment of this application, the population consists of multiple individuals, and the step of controlling the initial population to undergo iterative evolution based on the archived historical success experiences includes the following steps:

[0124] Generate the current fitness value evaluation count and the maximum fitness value evaluation count;

[0125] Before the iteration loop begins, the current fitness value evaluation count can be set to 0. The maximum number of fitness value evaluations serves as the termination condition for iterative evolution and can be generated based on the complexity of the actual problem, computational resource constraints, and algorithm characteristics; this application does not impose specific restrictions on this.

[0126] The initialized population is defined as the current population;

[0127] In the first iteration of evolution, a pre-generated initial population can be used as the current population for iterative evolution.

[0128] Based on the archived historical success experiences, the control parameters corresponding to the individuals in the current population are determined.

[0129] Among them, the archive of historical success experiences is associated with the corresponding individuals in the population.

[0130] The mutation vectors corresponding to the individuals in the current population are determined based on the control parameters.

[0131] The mutation vector is obtained by mutating individuals in the current population using the corresponding mutation operator.

[0132] The crossover vectors corresponding to the individuals in the contemporary population are determined based on the mutation vectors.

[0133] The crossover vector can be generated by applying a crossover strategy to the generated mutation vector. The crossover strategy can recombine the various dimensions of the mutation vector to generate a crossover vector with potentially better performance.

[0134] The fitness value of the crossover vector is evaluated to obtain the preferred vector of the current population;

[0135] Update the historical success experience archive and the current population according to the preferred vector;

[0136] The fitness value is calculated using the objective function Fitness and can be used to evaluate the quality of individuals in the corresponding population during iterative evolution.

[0137] Update the current fitness value evaluation count;

[0138] The current fitness value is evaluated 1 times, and the corresponding number is incremented by 1 with each update.

[0139] If the number of times the current fitness value is evaluated is less than the maximum number of times the fitness value is evaluated, return to the control parameters corresponding to the population individuals in the current population determined based on the archive of historical successful experiences;

[0140] If the current fitness value evaluation count equals the maximum fitness value evaluation count, output the result of the iterative evolution.

[0141] The current fitness value evaluation count and the maximum fitness value evaluation count can be pre-generated. At the start of the first iteration, the current fitness value evaluation count can be set to 0. The maximum fitness value evaluation count can be generated based on the complexity of the actual problem, computational resource limitations, and algorithm characteristics; this application does not impose specific restrictions on this. The initialized population can be determined as the current population to begin iterative evolution. Control parameters corresponding to individuals in the current population can be determined based on historical successful experience archives. Mutation vectors corresponding to individuals in the current population can be determined based on the control parameters, and crossover vectors corresponding to individuals in the current population can be determined based on the mutation vectors. The optimal vector for the current population is obtained by evaluating the fitness value of the crossover vectors, and then the historical successful experience archive and the current population are updated based on the optimal vector. The current fitness value evaluation count is updated, incrementing by 1 with each update. The current fitness value evaluation count is compared with the preset maximum fitness value evaluation count. If the current fitness value evaluation count is less than the maximum fitness value evaluation count, the control parameters corresponding to individuals in the current population determined based on historical successful experience archives are returned, and subsequent steps are executed to achieve iterative evolution. If the current fitness value evaluation count equals the maximum fitness value evaluation count, the iteration ends and the iterative evolution result is output. Based on the above implementation process, historical successful experiences can be used to control the initialized population to undergo adaptive iterative evolution in a better direction. This application directly identifies the second-level parameters of MF-Tyre based on improved L-SHADE. Compared to staged identification methods, this results in more robust and generalizable parameter results. Compared to methods for identifying first-level parameters, it can cover a wider range of engineering scenarios, such as the changing vertical load during vehicle operation. Compared to methods for identifying second-level parameters, it achieves higher identification accuracy.

[0142] In one embodiment of this application, the historical success experience archive includes the average crossover rate and average mutation factor corresponding to the individuals of the population that have successfully evolved, and the average crossover rate and average mutation factor preset when the historical success experience archive was generated. The step of determining the control parameters corresponding to the individuals of the current population based on the historical success experience archive includes:

[0143] The average crossover rate is randomly sampled to obtain the sampled crossover rate;

[0144] The adaptive crossover rate for each individual in the current generation of the population is generated based on the sampling crossover rate.

[0145] Random sampling is performed on the average mutation factor to obtain the sampled mutation factor;

[0146] The adaptive mutation factor for each individual in the current population is generated based on the sampled mutation factor; the control parameters include the adaptive crossover rate and the adaptive mutation factor.

[0147] When generating historical success experience archives, the corresponding individuals in the population have not yet begun iterative evolution. At this time, preset average crossover rates and average mutation factors can be generated in the historical success experience archives. For example, the preset average crossover rate (MCR) and average mutation factor (MF) can both be set to 0.5, and the size of each historical success experience archive can be a preset historical window size. After each subsequent generation iteration, the average crossover rate and average mutation factor adopted by the currently successfully evolved individuals in the population are calculated and saved to the historical success experience archives. The control parameters corresponding to the individuals in the current population include adaptive crossover rate and adaptive mutation factor. The average crossover rate stored in the historical success experience archives can be randomly sampled to obtain a sampled crossover rate, and the adaptive crossover rate corresponding to the individuals in the current population can be generated based on the sampled crossover rate. The average mutation factor in the historical success experience archives can be randomly sampled to obtain a sampled mutation factor, and the adaptive mutation factor corresponding to the individuals in the current population can be generated based on the sampled mutation factor. Based on the above implementation process, adaptive parameter adjustment can be achieved, and the mutation factor and crossover rate can be dynamically adjusted according to real-time feedback during the search process, thereby effectively adapting to the high complexity of the magic formula parameter space and its potential multi-peak characteristics.

[0148] For example, for the i-th individual in the current population, the corresponding control parameters can be generated according to the archive of historical success experiences associated with that individual, using the following formula:

[0149]

[0150] Where i = 1, 2, ..., N, N is the population size of the current generation; G is the current iteration number; CR is the adaptive crossover rate. CR represents the adaptive crossover rate of the i-th individual in the population during the G-th generation, where CR ∈ [0,1]; F is the adaptive mutation factor. F represents the adaptive mutation factor of the i-th individual in the population during the G-th generation, where F ∈ [0,1]; randn is a random value obtained based on normal distribution sampling, randc is a random value obtained based on Cauchy distribution sampling, and MCR k The sampled crossover rate is obtained by randomly sampling the saved average crossover rate based on index k in the historical successful experience archive; MF kThe sampled mutation factor is obtained by randomly sampling the average mutation factor stored in the historical successful experience archive according to index k. k = 1, ..., H, where H is the size of the historical successful experience archive. When the sampling crossover rate is equal to 1, the corresponding generated adaptive crossover rate is 0; otherwise, the corresponding generated adaptive crossover rate is based on a normal distribution (MCR). k The random value is obtained by random sampling of (0.1). The adaptive mutation factor is based on the Cauchy distribution (MF). k The random value obtained by random sampling of ,0.1).

[0151] For example, when generating a historical success experience archive, an external archive can also be generated. The generated external archive is set to an empty set and is used to store the inferior solutions of each generation of iterative evolution.

[0152] In one embodiment of this application, determining the mutation vector corresponding to an individual in the current population based on the control parameters includes:

[0153] The mutation operator to be used is determined based on the number of evaluations of the current fitness value at the start of the evolution of the contemporary population, and the mutation operator includes global mutation operators and local mutation operators;

[0154] Based on the control parameters and the mutation operator used, the individuals in the population are mutated to obtain the mutation vector.

[0155] Based on the generated control parameters, a mutation strategy can be executed on individuals in the current population to determine the corresponding mutation vector. The mutation operator to be used can be determined by evaluating the current fitness value at the start of the current population's evolution. Mutation operators include global mutation operators and local mutation operators. Based on the control parameters and the chosen mutation operator, mutations can be performed on individuals in the population to obtain mutation vectors.

[0156] For example, in the improved L-SHADE, two mutation operators, DE / current-to-pbest / 1 and DE / current-to-ord_pbest / 1, are used respectively. DE / current-to-pbest / 1 focuses more on global exploration, while DE / current-to-ord_pbest / 1 focuses more on local exploration. The two operators are as follows:

[0157] The formula corresponding to DE / current-to-pbest / 1 is:

[0158]

[0159] in, Here, G represents the mutation vector corresponding to an individual in the population, j represents the corresponding dimension, and x represents the mutation vector.i Let i be the solution vector corresponding to an individual in the population, where i represents the nth individual in the current population. The individuals were randomly selected from the top p% of the population with the best fitness values ​​in the current population, where p∈(0,1]. Randomly selected from the current population. The results are randomly selected from the union of the current population and external archives, where p≠r1≠r2, F i The mutation factor is the control parameter corresponding to an individual in the population.

[0160] The formula corresponding to DE / current-to-ord_pbest / 1 is:

[0161]

[0162] In DE / current-to-ord_pbest / 1, firstly, a vector is randomly selected from the top 100p% of individuals by fitness value. Then, two vectors are selected from the current population and from the union of the population and external archives, respectively. These vectors are required to be distinct from each other and not identical to i. They are then ranked by fitness value, and the three vectors are divided into x groups based on their fitness values ​​from best to worst. ord_pbest x ord_pmedian x ord_pworst . Here, G represents the mutation vector corresponding to an individual in the population, j represents the corresponding dimension, and x represents the mutation vector. i Let F be the solution vector corresponding to an individual in the population, where i represents the nth individual in the current population. i The mutation factor is the control parameter corresponding to an individual in the population.

[0163] For example, refer to Figure 4 The flowchart shown here illustrates a mutation strategy scheduling mechanism of an embodiment of the tire model parameter identification method of this application. It can evaluate the number of iterations based on the current fitness value at the start of the current population evolution and calculate the sampling probability P of each individual in the current population for the two mutation operators according to the following formula. G :

[0164]

[0165] Where randn is a random value obtained based on sampling from a normal distribution, nfes is the number of times the current fitness value is evaluated, and max_nfes is the maximum number of times the fitness value is evaluated.

[0166] For the i-th individual in the current population, generate a random number rand between 0 and 1 based on a uniform distribution, and determine whether the random number randi corresponding to each individual is less than the sampling probability P. G When randi is less than PG In the case of selecting the mutation operator DE / current-to-ord_pbest / 1, when randi is not less than P G In this case, the mutation operator DE / current-to-pbest / 1 is selected, and then the individuals in the current population are mutated based on the selected operator to obtain the corresponding mutation vector. According to the above implementation process, the sampling probability of the mutation strategy can be dynamically adjusted based on the number of evaluations to determine the current stage of the population. The current stage of the population includes the exploration stage and the development stage. When the population tends towards the exploration stage, the mutation strategy DE / current-to-pbest / 1, which is beneficial for global search, is sampled first. Conversely, when the population enters the development stage, the probability of sampling the mutation strategy DE / current-to-ord_pbest / 1, which is more suitable for local development, is increased. This achieves an adaptive balance between global exploration and local development capabilities at different evolutionary stages, thereby effectively improving the overall performance of the algorithm.

[0167] In one embodiment of this application, after determining the mutation vectors corresponding to individuals in the current population, individuals whose mutation vectors exceed the boundaries can be corrected. For example, the formula corresponding to the correction strategy is shown below:

[0168]

[0169] Where U is the upper boundary and L is the lower boundary. i Let x be the mutation vector corresponding to an individual in the current population. i Let j represent the population individuals at the start of the current population evolution, and G represent the corresponding dimension. Values ​​exceeding the boundary can be set to half the sum of the boundary value (including the upper or lower boundary) and the value of the corresponding population individuals at the start of the current population evolution. Similarly, if a dimension of a current population individual exceeds the lower boundary, the value of that dimension is corrected to half the sum of the lower boundary and the corresponding dimension value of the current population individual at the start of the current population evolution. Likewise, if a dimension of a current population individual exceeds the upper boundary, the value of that dimension is corrected to half the sum of the upper boundary and the corresponding dimension value of the current population individual at the start of the current population evolution.

[0170] In one embodiment of this application, determining the crossover vector corresponding to the individuals in the current population based on the mutation vector includes:

[0171] Based on the mutation vector and the current vector corresponding to the individual in the population at the start of the contemporary population evolution, determine each dimension of the crossover vector;

[0172] The cross vector is generated based on each dimension of the cross vector.

[0173] After executing the mutation strategy, a crossover strategy can be further executed. Based on the determined mutation vector and the current vector corresponding to an individual in the population at the start of the current population evolution, the dimensions of the crossover vector are determined, and then a crossover vector is generated based on these dimensions. The crossover strategy can employ binary crossover. Following this process, the obtained mutation vectors can be further exchanged and recombinated, which helps explore new regions in the search space and thus find better solutions.

[0174] For example, in the iterative evolution process of the Gth generation, the crossover vector corresponding to the mutation vector can be generated based on the binary crossover. Its d-th dimension Determined by the following formula:

[0175]

[0176] in, Let d be the d-th dimension of the corresponding mutation vector. Let be the d-th dimension of the corresponding current vector, where d = 1, 2, ..., D, and D is the dimension of the solution vector corresponding to an individual in the population. rand ∈ [0, 1] represents a uniformly distributed random number. For adaptive crossover rate, d rand The dimension is randomly selected to ensure that at least one dimension is replaced.

[0177] For the current vector x corresponding to an individual in the population at the beginning of the current population evolution... i and its mutation vector v i It can generate random numbers dimension by dimension, and when the random number is less than Or d = d rand In the case of a mutation vector, select the value of that dimension; otherwise, select the value of that dimension of the target vector. Generate a complete cross vector based on the values ​​of each dimension corresponding to the selected cross vector.

[0178] In one embodiment of this application, evaluating the fitness value of the crossover vector to obtain the preferred vector of the current population includes:

[0179] The selection strategy is implemented by comparing the crossover vectors of individuals in the current population with the fitness values ​​of the current vectors of individuals at the start of the current population evolution, based on the objective function. The vector with the smaller fitness value is retained as the preferred vector.

[0180] For example, the preferred vector can be determined using the following formula:

[0181]

[0182] in, This is the preferred vector for the current population. This represents the crossover vector corresponding to the individuals in the current population. The current vector of an individual in the population at the start of the current population evolution. The fitness value of the crossover vector is calculated based on the objective function. Let be the fitness value of the current vector calculated based on the objective function. i = 1, 2, ..., N, where N is the population size and G is the current iteration number. If the fitness value of the crossover vector is less than the fitness value of the current vector, the corresponding crossover vector is determined as the preferred vector; otherwise, the corresponding current vector is determined as the preferred vector.

[0183] In one embodiment of this application, updating the historical success experience archive and the current population according to the preferred vector includes:

[0184] The preferred vectors whose fitness values ​​are less than a preset threshold are identified as population individuals that have successfully evolved in the contemporary population.

[0185] Calculate the average crossover rate and average mutation factor for the individuals in the successfully evolved population.

[0186] The historical success experience archive is updated based on the average crossover rate and the average mutation factor.

[0187] The next generation of individuals and individuals to be eliminated are determined based on the fitness values ​​corresponding to the preferred vectors.

[0188] From the individuals in the population to be eliminated, select individuals with potential potential populations;

[0189] The current population is updated based on the potential population individuals and the next generation population individuals, wherein the potential population individuals are population individuals whose current fitness value is less than that of the previous generation and are to be eliminated.

[0190] The preset threshold corresponding to the fitness value is used to select individuals in the current population whose post-evolutionary fitness values ​​meet the requirements. Preferred vectors with fitness values ​​less than the preset threshold can be identified as successfully evolved individuals in the current population. The average crossover rate and average mutation factor corresponding to these successfully evolved individuals are calculated, and then the average crossover rate and average mutation factor in the historical success experience archive can be updated in pairs. Next-generation individuals and individuals to be eliminated can be determined based on the fitness values ​​corresponding to the preferred vectors. Potential individuals are then selected from the individuals to be eliminated, and the current population is updated based on these potential individuals and next-generation individuals, resulting in the current population used for next-generation iterative evolution. Potential individuals are those whose current fitness value is less than the previous generation's fitness value and are considered to be eliminated. The updated population includes both potential individuals and next-generation individuals. Based on the above implementation process, adaptive nonlinear population size reduction can be achieved according to the improvement of the fitness values ​​of individuals in the current population. When there is high diversity in the exploration phase, the reduction can be slowed down to retain global search capabilities, while when there is insufficient diversity in the development phase, the reduction can be accelerated to focus on high-quality solutions and ensure more efficient use of computing resources.

[0191] For example, after evolutionarily updating the current population, the historical successful experience archives MCR and MF of pairwise updates can be calculated according to the following formula:

[0192]

[0193] Here, SCR includes the crossover probability used by individuals that successfully evolved in the current population during this iteration, and SF includes the mutation factor used by individuals that successfully evolved in the current population during this iteration, k = 1, ..., H, where H is the size of the historical success experience archive. When SCR is empty, it means there are no successfully evolved individuals in the current population; in this case, the average crossover rate is updated to -1. When MCR... k When = -1, MCR kThe algorithm will enter a stagnation phase, forcing it to implement a "one parameter change at a time" strategy. This involves setting the crossover rate of the next generation of individuals to 0 to address some multimodal issues. If the maximum crossover rate stored in the SCR is 0, the average crossover rate is updated to the average crossover rate of the previous generation. If there are successfully evolved individuals in the current population, and some of these individuals have non-zero crossover rates, the average crossover rate is updated to the second-order weighted Lehmer average of the SCR. If the SF is empty, indicating no successfully evolved individuals in the current population, the average mutation factor is updated to the average mutation factor of the previous generation; otherwise, the mutation factor is updated to the second-order weighted Lehmer average of the SF. The updated average crossover rate and average mutation factor are then stored in pairs in the historical success experience archive.

[0194] mean WL (S) represents the second-order weighted Lehmer average, which can be calculated using the following formula:

[0195]

[0196] Here, S can refer to either SCR or SF. |S| is the number of crossover probabilities used by individuals that successfully evolved in the current iteration within the SCR, or the number of mutation factors used by individuals that successfully evolved in the current iteration within the SF, l = 1, ..., |S|. l To show the improvement in fitness value, s l This represents the crossover probability in SCR or the mutation factor in SF. The fitness value of the crossover vector of successfully evolved individuals in the population, calculated based on the objective function. Δf represents the fitness value of the current vector of a successfully evolved individual in the population, calculated based on the objective function. l It is the absolute value of the difference between the fitness value of the crossover vector of a successfully evolved individual in the population and the fitness value of the current vector.

[0197] In L-SHADE, the next generation population size N G+1 As the number of evaluations decreases, the size of the next generation population after the current population update can be calculated using the following formula, which is the number of individuals in the updated next generation population:

[0198]

[0199] The `round` function returns a rounded integer, `nfes` is the number of times the fitness value has been evaluated, `max_nfes` is the maximum number of times the fitness value has been evaluated, and N... MAX N represents the initial population size. MINThis represents the minimum allowed population size.

[0200] Reference Figure 5 The diagram illustrates an adaptive population size reduction flowchart of an embodiment of the tire model parameter identification method of this application. The population size for the next generation can be calculated, and then the individuals in the current population are sorted based on their fitness values, with individuals having lower fitness values ​​ranking higher. A number of individuals ranking high are selected as the next generation population based on the calculated population size, while the remaining individuals ranking lower are designated as individuals to be eliminated. For each eliminated individual, its fitness value is checked to see if it improves in the current generation. If the fitness value of the individual has improved, it is retained for the next evolution; otherwise, it is eliminated and stored in an external archive A. For example, if the size of the external archive exceeds the capacity limit, some individuals in the external archive can be randomly removed. Based on the above implementation process, an adaptive population size reduction mechanism based on fitness value improvement is adopted. When performing population size reduction operation, for individuals that are about to be eliminated, if their fitness value has improved in the current evolution, they are determined to still have optimization potential, and thus they are retained in the current population instead of being directly eliminated. This can dynamically balance exploration and development efficiency, and effectively reduce the waste of computing resources while avoiding premature convergence.

[0201] In one embodiment of this application, the target tire model parameters corresponding to the tire model can be determined based on the results of iterative evolution. After the population iterative evolution is completed, an evolved population is obtained. The fitness value of each individual in the evolved population can be calculated, and the individual with the smallest fitness value is selected as the target individual. The target tire model parameters corresponding to the tire model can be determined based on the values ​​of each dimension of the solution vector corresponding to the target individual.

[0202] For example, refer to Figure 6The diagram illustrates the identification effect of an embodiment of the tire model parameter identification method of this application. Taking a certain brand and model 185 / 65R15 tire as an example, the curves in the diagram represent the slip angle-lateral force curves generated based on the verified optimal parameters, and the dots represent the fitting points of the slip angle-lateral force data obtained from the target tire model parameters obtained using the tire model parameter identification method of this application. The curves in the diagram correspond to the 1st to 8th vertical loads according to the degree of curvature from smallest to largest. The curves and dots corresponding to the same vertical load are respectively colored. It can be seen that the tire model parameter identification method of this application has high identification accuracy. The fitting results of the slip angle-lateral force data corresponding to the target tire model parameters obtained by this method highly overlap with the slip angle-lateral force curve generated by the verified optimal parameters.

[0203] Reference Figure 7 This is a flowchart illustrating the steps of another embodiment of the tire model parameter identification method of this application. After starting tire model parameter identification, a tire dynamic characteristic test bench equipped with lateral force sensors, longitudinal force sensors, and self-aligning torque sensors can be used. By inputting the tire slip angle and longitudinal slip ratio, the lateral force, longitudinal force, and self-aligning torque output by the test bench can be obtained. The obtained data such as tire slip angle, longitudinal slip ratio, lateral force, longitudinal force, and self-aligning torque are then processed. For example, unit conversion, outlier removal, and negative augmentation can be performed on the obtained data.

[0204] Then, an initialized population can be generated as the current generation population. It is then determined whether the current generation population meets the termination condition. The termination condition can be set to the current fitness value evaluation count being equal to the preset maximum fitness value evaluation count. If the termination condition is not met, the crossover rate (SCR) and mutation factor (SF) used by the individuals that successfully evolved in the current generation population in this iteration are used to generate a historical success experience archive. Based on this archive, control parameters for each individual are generated, including the crossover rate (CR) and mutation factor (F).

[0205] A mutation strategy scheduling mechanism is adopted. Based on the number of evaluations of the current fitness value at the beginning of the current generation, the sampling probability of each individual in the current population for the two mutation operators DE / current-to-pbest / 1 and DE / current-to-ord_pbest / 1 is calculated. Then, a random number rand between 0 and 1 is generated for each individual based on a uniform distribution. If the random number corresponding to the individual is less than the sampling probability, DE / current-to-ord_pbest / 1 is selected as the mutation operator. If the random number corresponding to the individual is not less than the sampling probability, DE / current-to-pbest / 1 is selected as the mutation operator. Then, each individual is mutated based on the selected operator to obtain the mutation vector corresponding to each individual.

[0206] A crossover strategy is adopted, and crossover vectors corresponding to each mutation vector of the current population individual are generated based on binary crossover.

[0207] A selection strategy is adopted, which compares the crossover vectors of individuals in the current population with the fitness values ​​of the current vectors of individuals at the start of the current population evolution based on the objective function, and retains the better value to obtain the preferred vector.

[0208] Record individual success information and update the average crossover rate (MCR) and average mutation factor (MF) in the historical success experience archive.

[0209] An adaptive population size reduction strategy is adopted. After calculating the population size corresponding to the next generation, the corresponding individuals to be eliminated are selected according to the fitness value of the preferred vector. It is determined whether the fitness value of each individual to be eliminated has improved in the current generation. If it has, the individual is retained to enter the next generation of evolution. If not, the individual is eliminated.

[0210] Update external save file A to store eliminated individuals from the population into the external save file.

[0211] Determine if the termination condition is met. If the termination condition is met, output the corresponding optimal solution. If the termination condition is not met, continue iteratively evolving the initialized population until the termination condition is met.

[0212] It should be noted that, for the sake of simplicity, the method embodiments are all described as a series of actions. However, those skilled in the art should understand that the embodiments of this application are not limited to the described order of actions, because according to the embodiments of this application, some steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also understand that the embodiments described in the specification are all preferred embodiments, and the actions involved are not necessarily required by the embodiments of this application.

[0213] Based on the above embodiments, this embodiment also provides a tire model parameter identification device, which can be applied to electronic devices such as terminal devices and servers.

[0214] Reference Figure 8 The diagram shows a structural block diagram of an embodiment of a tire model parameter identification device according to this application, which may specifically include the following modules:

[0215] The acquisition module 801 is used to acquire the parameters of the tire model to be identified.

[0216] Population generation module 802 is used to generate an initialized population based on the parameters of the tire model to be identified;

[0217] The historical success experience archive generation module 803 is used to generate a historical success experience archive; the historical success experience archive is used to control the evolutionary direction of the initialized population.

[0218] The iterative evolution module 804 is used to control the initial population to undergo iterative evolution based on the archived historical success experience.

[0219] The determination module 805 is used to determine the target tire model parameters corresponding to the tire model based on the results of the iterative evolution.

[0220] The population generation module is further used for:

[0221] Based on the parameters of the tire model to be identified, a preset number of uniformly distributed random values ​​are selected from the preset search space as the population individuals; the dimension of the population individuals corresponds to the number of parameters of the tire model to be identified; the initialized population includes a preset number of population individuals.

[0222] The iterative evolution module includes:

[0223] The generation submodule is used to generate the current fitness value evaluation count and the maximum fitness value evaluation count;

[0224] The contemporary population determination submodule is used to determine the initialized population as the contemporary population; the population consists of multiple population individuals;

[0225] The control parameter determination submodule is used to determine the control parameters corresponding to the individuals in the current population based on the archived historical success experience.

[0226] The mutation vector determination submodule is used to determine the mutation vector corresponding to the population individuals of the current population based on the control parameters.

[0227] The crossover vector determination submodule is used to determine the crossover vector corresponding to the individuals in the current population based on the mutation vector;

[0228] An evaluation submodule is used to evaluate the fitness value of the crossover vector to obtain the preferred vector of the current population;

[0229] The first update submodule is used to update the historical success experience archive and the current population according to the preferred vector;

[0230] The second update submodule is used to update the current fitness value evaluation count;

[0231] The return submodule is used to return the control parameters corresponding to the population individuals of the current population determined based on the historical successful experience archive when the number of current fitness value evaluations is less than the maximum number of fitness value evaluations.

[0232] The output submodule is used to output the results of iterative evolution when the number of times the current fitness value is evaluated is equal to the maximum number of times the fitness value is evaluated.

[0233] The historical success experience archive includes the average crossover rate and average mutation factor corresponding to the individuals in the population that have successfully evolved, as well as the average crossover rate and average mutation factor preset when the historical success experience archive was generated. The control parameter determination submodule includes:

[0234] A crossover rate sampling unit is used to randomly sample the average crossover rate to obtain a sampled crossover rate.

[0235] An adaptive crossover rate generation unit is used to generate an adaptive crossover rate for each individual in the current generation of the population based on the sampled crossover rate.

[0236] A mutation factor sampling unit is used to randomly sample the average mutation factor to obtain a sampled mutation factor.

[0237] An adaptive mutation factor generation unit is used to generate an adaptive mutation factor for the current population individuals based on the sampled mutation factor; the control parameters include the adaptive crossover rate and the adaptive mutation factor.

[0238] The mutation vector determination submodule includes:

[0239] The mutation operator determination unit is used to determine the mutation operator to be used based on the number of evaluations of the current fitness value at the start of the evolution of the contemporary population. The mutation operator includes a global mutation operator and a local mutation operator.

[0240] A mutation unit is used to mutate individuals in the population based on the control parameters and the mutation operator used, to obtain the mutation vector.

[0241] The cross vector determination submodule is further used for:

[0242] Based on the mutation vector and the current vector corresponding to the individual in the population at the start of the contemporary population evolution, determine each dimension of the crossover vector;

[0243] The cross vector is generated based on each dimension of the cross vector.

[0244] The first update submodule is further used for:

[0245] The preferred vectors whose fitness values ​​are less than a preset threshold are identified as population individuals that have successfully evolved in the contemporary population.

[0246] Calculate the average crossover rate and average mutation factor for the individuals in the successfully evolved population.

[0247] The historical success experience archive is updated based on the average crossover rate and the average mutation factor.

[0248] The next generation of individuals and individuals to be eliminated are determined based on the fitness values ​​corresponding to the preferred vectors.

[0249] From the individuals in the population to be eliminated, select individuals with potential potential populations;

[0250] The current population is updated based on the potential population individuals and the next generation population individuals, wherein the potential population individuals are population individuals whose current fitness value is less than that of the previous generation and are to be eliminated.

[0251] This application provides a tire model parameter identification device that can acquire the tire model parameters to be identified, generate an initial population based on the tire model parameters, generate a historical success experience archive to control the evolutionary direction of the initial population, and then control the initial population to iteratively evolve based on the historical success experience archive. Based on the results of the iterative evolution, the target tire model parameters corresponding to the tire model are determined, achieving high-precision identification of the tire model parameters corresponding to the tire model. The generated historical success experience archive allows for adaptive adjustment during the evolutionary process, resulting in better generalization ability and robustness of the final identification result, thereby improving the accuracy of the corresponding tire model.

[0252] As the device embodiment is basically similar to the method embodiment, the description is relatively simple, and relevant parts can be found in the description of the method embodiment.

[0253] This application provides one or more electronic devices, including: a processor; and a memory storing executable code thereon, which, when executed, causes the processor to perform one or more of the methods described in the above embodiments.

[0254] This application provides one or more machine-readable media storing instructions that, when executed by one or more processors, cause an electronic device to perform one or more of the methods described in the above embodiments. In this application, the electronic device includes various types of devices such as terminal devices and servers (clusters).

[0255] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0256] This application describes embodiments with reference to flowchart illustrations and / or block diagrams of methods, terminal devices (systems), and computer program products according to embodiments of this application. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable tire model parameter identification terminal device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable tire model parameter identification terminal device, generate instructions for implementing the process... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0257] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable tire model parameter identification terminal device to operate in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0258] These computer program instructions can also be loaded onto a computer or other programmable tire model parameter identification terminal device, causing a series of operational steps to be executed on the computer or other programmable terminal device to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable terminal device for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0259] Although preferred embodiments of the present application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of the present application.

[0260] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes said element.

[0261] The above provides a detailed description of a tire model parameter identification method and apparatus, an electronic device, and a storage medium provided in this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for identifying tire model parameters, characterized in that, The method includes: Obtain the parameters of the tire model to be identified; An initial population is generated based on the parameters of the tire model to be identified; Generate a historical success experience archive; the historical success experience archive is used to control the evolutionary direction of the initialized population; Based on the archived historical successes, the initialized population is controlled to undergo iterative evolution. The target tire model parameters corresponding to the tire model are determined based on the results of the iterative evolution. The method further includes: The initialized population is defined as the current population; The sampling probabilities of each individual in the current population for the global mutation operator and the local mutation operator are calculated based on the number of evaluations of the current fitness value at the start of the evolution of the current population. A random number is generated for each individual. If the random number is less than the sampling probability, the local mutation operator is selected; otherwise, the global mutation operator is selected. The individuals are mutated based on the selected mutation operator to obtain the mutation vector corresponding to each individual. The current population is then iteratively evolved based on the mutation vector.

2. The method according to claim 1, characterized in that, The process of generating an initialized population based on the parameters of the tire model to be identified includes: Based on the parameters of the tire model to be identified, a preset number of uniformly distributed random values ​​are selected from the preset search space as the population individuals; the dimension of the population individuals corresponds to the number of parameters of the tire model to be identified; the initialized population includes a preset number of population individuals.

3. The method according to claim 1, characterized in that, The population consists of multiple individuals, and the iterative evolution of the initialized population based on the archived historical success experiences includes the following steps: Generate the current fitness value evaluation count and the maximum fitness value evaluation count; Based on the archived historical success experiences, the control parameters corresponding to the individuals in the current population are determined. The mutation vectors corresponding to the individuals in the current population are determined based on the control parameters. The crossover vectors corresponding to the individuals in the contemporary population are determined based on the mutation vectors. The fitness value of the crossover vector is evaluated to obtain the preferred vector of the current population; Update the historical success experience archive and the current population according to the preferred vector; Update the current fitness value evaluation count; If the number of times the current fitness value is evaluated is less than the maximum number of times the fitness value is evaluated, return to the control parameters corresponding to the population individuals in the current population determined based on the archive of historical successful experiences; If the current fitness value evaluation count equals the maximum fitness value evaluation count, output the result of the iterative evolution.

4. The method according to claim 3, characterized in that, The historical success experience archive includes the average crossover rate and average mutation factor corresponding to the individuals of the population that have successfully evolved, as well as the average crossover rate and average mutation factor preset when the historical success experience archive was generated. The step of determining the control parameters corresponding to the individuals of the current population based on the historical success experience archive includes: The average crossover rate is randomly sampled to obtain the sampled crossover rate; The adaptive crossover rate for each individual in the current generation of the population is generated based on the sampling crossover rate. Random sampling is performed on the average mutation factor to obtain the sampled mutation factor; The adaptive mutation factor for each individual in the current population is generated based on the sampled mutation factor; the control parameters include the adaptive crossover rate and the adaptive mutation factor.

5. The method according to claim 3, characterized in that, Determining the mutation vector corresponding to an individual in the current population based on the control parameters includes: The mutation operator to be used is determined based on the number of evaluations of the current fitness value at the start of the evolution of the contemporary population, and the mutation operator includes global mutation operators and local mutation operators; Based on the control parameters and the mutation operator used, the individuals in the population are mutated to obtain the mutation vector.

6. The method according to claim 3, characterized in that, The step of determining the crossover vector corresponding to the individuals in the current population based on the mutation vector includes: Based on the mutation vector and the current vector corresponding to the individual in the population at the start of the contemporary population evolution, determine each dimension of the crossover vector; The cross vector is generated based on each dimension of the cross vector.

7. The method according to claim 3, characterized in that, The step of updating the historical success experience archive and the current population according to the preferred vector includes: The preferred vectors whose fitness values ​​are less than a preset threshold are identified as population individuals that have successfully evolved in the contemporary population. Calculate the average crossover rate and average mutation factor for the individuals in the successfully evolved population. The historical success experience archive is updated based on the average crossover rate and the average mutation factor. The next generation of individuals and individuals to be eliminated are determined based on the fitness values ​​corresponding to the preferred vectors. From the individuals in the population to be eliminated, select individuals with potential potential populations; The current population is updated based on the potential population individuals and the next generation population individuals, wherein the potential population individuals are population individuals whose current fitness value is less than that of the previous generation and are to be eliminated.

8. A tire model parameter identification device, characterized in that, The device includes: The acquisition module is used to acquire the parameters of the tire model to be identified. The population generation module is used to generate an initial population based on the parameters of the tire model to be identified. A historical success experience archive generation module is used to generate a historical success experience archive; the historical success experience archive is used to control the evolutionary direction of the initialized population; The iterative evolution module is used to control the initial population to undergo iterative evolution based on the archived historical success experiences. The determination module is used to determine the target tire model parameters corresponding to the tire model based on the results of the iterative evolution. The apparatus is further configured to: determine the initialized population as the current population; calculate the sampling probability of each individual in the current population for the global mutation operator and the local mutation operator based on the number of evaluations of the current fitness value at the start of the evolution of the current population; generate a random number for each individual in the population; if the random number is less than the sampling probability, select the local mutation operator; otherwise, select the global mutation operator; mutate the individuals in the population based on the selected mutation operator to obtain the mutation vector corresponding to each individual in the population; and iteratively evolve the current population based on the mutation vector.

9. An electronic device, characterized in that, include: processor; and A memory having executable code stored thereon, which, when executed, causes the processor to perform the tire model parameter identification method as described in one or more of claims 1-7.

10. A computer-readable storage medium having executable code stored thereon, which, when executed, causes a processor to perform the tire model parameter identification method as described in one or more of claims 1-7.

Images