Dynamic model construction method, vehicle and storage medium

Abstract

The embodiment of the invention provides a dynamic model construction method, a vehicle and a storage medium. The method comprises the steps that feature parameters of vehicle tires on multiple feature dimensions are acquired; coding the feature parameters based on the parameter types of the feature parameters to obtain feature vectors of the feature parameters; predicting dynamic characteristics of the vehicle tire based on the feature vector to obtain magic parameters of the vehicle tire; and constructing a dynamic model of the vehicle tire based on the magic parameters. According to the method and the device, the technical problem of relatively low construction efficiency of the dynamic model of the vehicle tire in related technologies is solved.

Description

Technical Field

[0001] This application relates to the field of tire dynamics, and more specifically, to a method for constructing a dynamic model, a vehicle, and a storage medium. Background Technology

[0002] As a key component in the interaction between a vehicle and the ground, the dynamic performance of tires directly affects the vehicle's handling stability, driving safety, and ride comfort. The accuracy of tire models is crucial in vehicle dynamics simulation and control system design. However, current methods for constructing tire dynamic models primarily rely on bench tests, resulting in low efficiency in model building.

[0003] There is currently no good solution to the above problems. Summary of the Invention

[0004] This application provides a method for constructing a dynamic model, a vehicle, and a storage medium to at least solve the technical problem of low efficiency in constructing dynamic models of vehicle tires in related technologies.

[0005] According to one aspect of the embodiments of this application, a method for constructing a dynamic model is provided, comprising: obtaining feature parameters of a vehicle tire in multiple feature dimensions; encoding the feature parameters based on the parameter type of the feature parameters to obtain feature vectors of the feature parameters; predicting the dynamic characteristics of the vehicle tire based on the feature vectors to obtain magic parameters of the vehicle tire; and constructing a dynamic model of the vehicle tire based on the magic parameters.

[0006] Furthermore, the dynamic characteristics of vehicle tires are predicted based on feature vectors to obtain the magic parameters of vehicle tires. This includes: inputting feature vectors into a parameter prediction model, using the parameter prediction model to predict the dynamic characteristics of vehicle tires under different operating conditions, and obtaining magic parameters. The operating conditions are used to characterize the mechanical behavior of vehicle tires under different stress conditions.

[0007] Furthermore, the parameter prediction model includes a feature extraction layer and multiple parameter prediction layers, with different parameter prediction layers corresponding to different operating conditions. The feature vector is input into the parameter prediction model, which is then used to predict the dynamic characteristics of the vehicle tire under different operating conditions to obtain magic parameters. This includes: using the feature extraction layer to extract features from the feature vector to obtain shared features, where the shared features characterize features used in predicting the dynamic characteristics of the vehicle tire under any operating condition; and using multiple parameter prediction layers to process the shared features to obtain the magic parameters of the vehicle tire under different operating conditions.

[0008] Furthermore, multiple parameter prediction layers are used to process the shared features to obtain the magic parameters of the vehicle tires under different operating conditions. This includes: during the processing of the shared features using any parameter prediction layer, obtaining the degree of correlation between the operating condition corresponding to the parameter prediction layer and any shared feature; based on the degree of correlation, selecting a target feature from the shared features, wherein the degree of correlation corresponding to the target feature is greater than the degree of correlation corresponding to other features in the shared features besides the target feature; and processing the target feature based on the parameter prediction layer to obtain the magic parameters of the vehicle tires under the operating conditions corresponding to the parameter prediction layer.

[0009] Furthermore, the operating conditions include: longitudinal force condition, lateral force condition, and self-centering torque condition. The longitudinal force condition characterizes the mechanical behavior of the vehicle tires during straight-line driving; the lateral force condition characterizes the mechanical behavior of the vehicle tires during steering; and the self-centering torque condition characterizes the mechanical behavior of the vehicle tires during the process of resuming straight-line driving after steering. Multiple parameter prediction layers include: a longitudinal force parameter prediction layer, a lateral force parameter prediction layer, and a self-centering torque prediction layer. Based on the parameter prediction layers, the target features are processed to obtain... The magic parameters for the vehicle tires under the corresponding operating conditions in the parameter prediction layer include: processing the target features based on the longitudinal force parameter prediction layer to obtain the longitudinal force magic parameters of the vehicle tires under the longitudinal force condition; processing the target features based on the lateral force parameter prediction layer to obtain the lateral force magic parameters of the vehicle tires under the lateral force condition; processing the target features based on the self-aligning torque parameter prediction layer to obtain the self-aligning torque magic parameters of the vehicle tires under the self-aligning torque condition; and obtaining the magic parameters based on the longitudinal force magic parameters, lateral force magic parameters, and self-aligning torque magic parameters.

[0010] Furthermore, the feature parameters are encoded based on their parameter types to obtain feature vectors, including: obtaining the parameter padding method and parameter encoding method corresponding to the feature parameters based on their parameter types; padding the feature parameters based on the parameter padding method to obtain padding parameters; performing outlier detection on the padding parameters to obtain outlier detection results; and encoding the padding parameters using the parameter encoding method when the outlier detection results indicate that there are no outliers in the padding parameters to obtain feature vectors.

[0011] Furthermore, the parameter types include: numerical parameters and categorical parameters. Numerical parameters are used to characterize that the feature parameters are numerically meaningful, while categorical parameters are used to characterize that the feature parameters are physical properties of vehicle tires. The feature parameters are filled based on the parameter filling method to obtain filled parameters, including: when the parameter type is numerical, interpolating the feature parameters to obtain filled parameters; when the parameter type is categorical, performing mode filling on the feature parameters to obtain filled parameters.

[0012] Furthermore, the filler parameters are encoded using a parametric encoding method to obtain the feature vector, including: when the parameter type is a classification parameter, the feature parameters are one-hot encoded to obtain the feature vector.

[0013] According to another aspect of the embodiments of this application, a dynamic model construction apparatus is also provided, comprising: a first acquisition module, configured to acquire feature parameters of a vehicle tire in multiple feature dimensions; a first encoding module, configured to encode the feature parameters based on the parameter type of the feature parameters to obtain feature vectors of the feature parameters; a first prediction module, configured to predict the dynamic characteristics of the vehicle tire based on the feature vectors to obtain magic parameters of the vehicle tire; and a first construction module, configured to construct a dynamic model of the vehicle tire based on the magic parameters.

[0014] According to another aspect of the embodiments of this application, a vehicle is also provided, including: a memory storing an executable program; and a processor for running the program, wherein the program executes the methods in various embodiments of this application when it runs.

[0015] According to another aspect of the embodiments of this application, a computer-readable storage medium is also provided, the computer-readable storage medium including a stored executable program, wherein, when the executable program is running, it controls the device where the computer-readable storage medium is located to perform the methods of various embodiments of this application.

[0016] According to another aspect of the embodiments of this application, a computer program product is also provided, including a computer program that, when executed by a processor, implements the methods of various embodiments of this application.

[0017] According to another aspect of the embodiments of this application, a computer program product is also provided, including a non-volatile computer-readable storage medium storing a computer program that, when executed by a processor, implements the methods in various embodiments of this application.

[0018] According to another aspect of the embodiments of this application, a computer program is also provided, which, when executed by a processor, implements the methods of the various embodiments of this application.

[0019] In this embodiment, the method involves acquiring feature parameters of a vehicle tire across multiple feature dimensions; encoding the feature parameters based on their parameter types to obtain feature vectors; predicting the dynamic characteristics of the vehicle tire based on these feature vectors to obtain magic parameters; and constructing a dynamic model of the vehicle tire based on these magic parameters. By comprehensively considering the multidimensional features of the vehicle tire, more comprehensive feature parameters can be obtained, providing richer and more accurate data input for magic parameter prediction. Further encoding of the feature parameters converts their format into a digital format suitable for magic parameter prediction, making the prediction of magic parameters more efficient and accurate. This reduces the time required to construct a dynamic model based on magic parameters, achieving the goal of rapidly constructing a dynamic model of the vehicle tire. This improves the efficiency of constructing a dynamic model of the vehicle tire, thus solving the technical problem of low efficiency in constructing dynamic models of vehicle tires in related technologies. Attached Figure Description

[0020] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0021] Figure 1 This is a flowchart of a method for constructing a dynamic model according to an embodiment of this application;

[0022] Figure 2 This is a schematic diagram of an optional tire multidimensional feature parameter system framework according to an embodiment of this application;

[0023] Figure 3 This is a schematic diagram of an optional prediction process for tire magic parameters based on multidimensional features according to an embodiment of this application;

[0024] Figure 4 This is a schematic diagram of an optional tire feature-parameter mapping database construction process according to an embodiment of this application;

[0025] Figure 5 This is a schematic diagram of an optional parameter prediction model according to an embodiment of this application;

[0026] Figure 6 This is a schematic diagram of the training process of an optional parameter prediction model according to an embodiment of this application;

[0027] Figure 7 This is a schematic diagram of an optional parameter processing procedure according to an embodiment of this application;

[0028] Figure 8 This is a schematic diagram of a dynamic model construction apparatus according to an embodiment of this application. Detailed Implementation

[0029] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0030] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0031] According to an embodiment of this application, an embodiment of a method for constructing a dynamic model is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0032] This embodiment provides a method for constructing a dynamic model. Figure 1 This is a flowchart of a method for constructing a dynamic model according to an embodiment of this application, such as... Figure 1 As shown, the process includes the following steps:

[0033] Step S102: Obtain the feature parameters of the vehicle tires in multiple feature dimensions.

[0034] The aforementioned feature dimensions can be parameters used to describe different aspects or attribute categories of tire characteristics, including but not limited to: geometric feature dimensions, structural feature dimensions, and performance index dimensions. Among them, geometric feature dimensions mainly focus on the tire's shape and size characteristics, while structural feature dimensions mainly focus on the design of the tire's internal structure, and performance index dimensions mainly focus on the tire's performance under specific usage conditions.

[0035] The aforementioned characteristic parameters can be quantitative indicators that specifically measure tire characteristics under the aforementioned characteristic dimensions.

[0036] In one optional embodiment, considering that tire performance is affected by multi-dimensional parameters such as geometric features, structural features, material features, and performance indicators, collecting multi-dimensional feature parameters can more comprehensively describe the physical properties of the tire, thereby providing richer reference information for the subsequent construction of the dynamic model. Based on this, the dynamic model construction system (hereinafter referred to as the construction system) can acquire the feature parameters of the vehicle tire in multiple feature dimensions through various different acquisition methods.

[0037] For ease of understanding, Figure 2 This is a schematic diagram of an optional tire multidimensional feature parameter system framework according to an embodiment of this application, as shown below. Figure 2 As shown, the multidimensional characteristic parameter system of tires includes ensemble characteristic dimension, structural characteristic dimension, and performance index dimension. Specifically, in order to accurately obtain the geometric characteristic dimension parameters of vehicle tires, the dimensions of the vehicle tire can be directly measured using measuring tools such as tape measures and measuring instruments when the vehicle tire is in a normal inflation state, thereby obtaining the tire cross-sectional width. The tire aspect ratio can also be read directly from the tire carcass, and the rim diameter on which the vehicle tire is mounted can be directly measured using measuring tools.

[0038] To accurately obtain parameters of the structural features of vehicle tires, tread features such as tread depth and tread type can be obtained through visual inspection or 3D scanning technology. Alternatively, the internal structure of the tire can be analyzed through technical manuals provided by the tire manufacturer or through X-ray imaging technology to obtain the tread ply structure, such as the number of plies, the material and arrangement of the plies, and the sidewall ply structure can be obtained through technical manuals or non-destructive testing technology.

[0039] To accurately obtain performance parameters of vehicle tires, the load-bearing capacity of the tires can be measured under static conditions using standard weighing equipment to obtain the vertical load. Alternatively, the internal tire pressure can be directly read using a tire barometer, and the tire wear index can be evaluated using wear indicators on the tire. Furthermore, the traction performance of vehicle tires can be determined on a tire test bench by simulating different road conditions and load states, thus obtaining the traction rating of the vehicle tires. Finally, the temperature rating that vehicle tires can withstand at high speeds can be obtained through thermal cycling tests in the laboratory.

[0040] Step S104: Encode the feature parameters based on their parameter types to obtain the feature vectors of the feature parameters.

[0041] The parameter types mentioned above can refer to the data category or attribute type to which the feature parameters belong. For example, they can be divided into numerical parameters and categorical parameters.

[0042] The aforementioned feature vector can be a series of values ​​formed by encoding the parameter types of the tire, which are then input into a deep neural network for prediction. Each dimension of the feature vector can correspond to a specific parameter type, such as the tire's cross-sectional width, tire pressure, tread design encoding value, etc., to ensure that the tire's multidimensional physical characteristics are converted into numerical forms that the machine learning model can understand, thereby enabling effective learning and prediction.

[0043] In one alternative embodiment, considering that dynamic models typically only need to process numerical data, and that tire feature parameters include various types of parameters, such as numerical parameters (e.g., tire cross-sectional width) and categorical feature parameters (e.g., tread pattern type), where numerical parameters can be directly used as input to the dynamic model, but the dynamic model may have difficulty directly processing categorical feature parameters, the system needs to pre-convert the categorical feature parameters into a numerical form that the dynamic model can understand, thereby improving the output efficiency and accuracy of the dynamic model. Based on this, the system can encode the feature parameters according to their parameter types to obtain feature vectors.

[0044] Specifically, for numerical parameters, different feature parameters may have different dimensions and units. In order to ensure that all feature parameters are compared on the same scale and to prevent some feature parameters from dominating the training process of the dynamic model due to their large numerical range, the system can use a standardization method to transform these feature parameters into a distribution with a mean of 0 and a standard deviation of 1, so as to eliminate the influence of dimensions and numerical range and improve the convergence speed and stability of the dynamic model. Alternatively, a normalization method can be used to map these feature parameter values ​​to the interval [0, 1], which also helps in the construction of the dynamic model.

[0045] The original classification parameters are discontinuous and unordered for the dynamic model, making it difficult for the model to directly understand the relationship between "category A" and "category B". Therefore, the system can use encoding transformation to convert these feature parameters into a format that the dynamic model can understand and manipulate. For example, to enable the dynamic model to better understand and process classification parameters, thereby improving prediction performance, the system can use one-hot encoding to convert a classification parameter into multiple binary features, each corresponding to a category. This makes the originally nonlinear classification parameter linearly separable, thus helping the dynamic model learn the importance of each classification parameter.

[0046] Step S106: Predict the dynamic characteristics of the vehicle tires based on the feature vectors to obtain the magic parameters of the vehicle tires.

[0047] The aforementioned dynamic characteristics can be the properties of the vehicle tire's motion state under the action of force, including but not limited to: the tire's force and torque response under different working conditions such as different loads, speeds, and road conditions. Specifically, they can be subdivided into longitudinal force characteristics that affect vehicle acceleration and braking, lateral force characteristics that affect vehicle turning and stability, and self-centering torque characteristics that affect the vehicle's self-centering ability. These characteristics together determine the vehicle tire's handling performance, safety, and comfort.

[0048] The aforementioned magic parameters can be parameters used to describe the mechanical behavior of vehicle tires, including but not limited to: longitudinal stiffness parameters, lateral stiffness parameters, self-aligning stiffness parameters, etc. By adjusting the magic parameters, the force and torque output of vehicle tires under different working conditions can be fitted, which are key parameters in vehicle dynamics simulation and control system design.

[0049] In one alternative embodiment, considering that the magic formula is a widely used semi-empirical model that can effectively describe these dynamic characteristics of tires, and that the accuracy of the magic formula is highly dependent on the precision of the magic parameters, in order to obtain magic parameters with high accuracy and thus improve the overall accuracy of the vehicle dynamics model, the system can first determine the mapping relationship between the multidimensional features of the vehicle tire and the magic parameters, and based on the feature vector, predict the dynamic characteristics of the vehicle tire according to the mapping relationship, thereby obtaining the magic parameters of the vehicle tire, so that the predicted magic parameters are closer to the dynamic behavior of the real tire.

[0050] In another alternative embodiment, to quickly and accurately assess the dynamic characteristics of tires under different operating conditions, the system can also use techniques such as multilayer perceptrons, convolutional neural networks, or recurrent neural networks to pre-build a deep learning model to learn the complex relationship between the feature vectors of the vehicle tires and the magic parameters. In practical applications, the system can input the feature vectors of the vehicle tires into the trained deep learning model, thereby using the model to predict the aforementioned magic parameters.

[0051] For ease of understanding, Figure 3 This is a schematic diagram illustrating an optional prediction process for tire magic parameters based on multidimensional features according to an embodiment of this application, such as... Figure 3As shown, the system can first construct a multi-dimensional feature parameter system for tires, and then build a tire feature-parameter mapping database based on this system. Subsequently, the system performs data preprocessing and feature engineering, such as missing value imputation, outlier detection and correction, and feature encoding, to improve data quality. Then, the system can use a deep neural network to construct an input layer, a shared feature extraction layer, and a task-specific prediction layer, and use the constructed model to predict the aforementioned magic parameters.

[0052] Figure 4 This is a schematic diagram of an optional tire feature-parameter mapping database construction process according to an embodiment of this application, such as... Figure 4 As shown, the system first establishes a standardized data acquisition process. Data sources can be divided into two parts: measured data from tire manufacturers, testing institutions, and research institutions, and simulation data collected using finite element tire models. Subsequently, the system can establish data quality control standards to clean the collected data, removing non-compliant and outlier data to ensure the purity and reliability of the database. Then, based on the cleaned data, the system can construct a tire feature-parameter mapping database to facilitate subsequent data analysis and model training.

[0053] Step S108: Based on the magic parameters, construct a dynamic model of the vehicle tires.

[0054] The aforementioned dynamic model can be used to predict the forces and moments generated by the tire under various working conditions, and then output key dynamic parameters such as the lateral force, longitudinal force, self-aligning torque, and roll torque of the vehicle tire.

[0055] In an alternative embodiment, considering that the aforementioned magic parameters provide a standardized way of describing the dynamic characteristics of vehicle tires, enabling data obtained from different vehicle tire tests to be compared and analyzed within the same framework, the construction of a dynamic model of vehicle tires based on magic parameters can make the dynamic model widely applicable to various tire types and operating conditions, thereby improving the applicability of the dynamic model.

[0056] Furthermore, the dynamic characteristics of vehicle tires are predicted based on feature vectors to obtain the magic parameters of vehicle tires. This includes: inputting feature vectors into a parameter prediction model, using the parameter prediction model to predict the dynamic characteristics of vehicle tires under different operating conditions, and obtaining magic parameters. The operating conditions are used to characterize the mechanical behavior of vehicle tires under different stress conditions.

[0057] The aforementioned parameter prediction model can be a model designed based on deep neural networks for predicting the magic parameters of vehicle tires. By learning the complex nonlinear mapping relationship between the multidimensional features of vehicle tires and the magic parameters, this model can achieve intelligent prediction of the dynamic behavior of tires under different working conditions, thereby assisting in the construction of a system to accurately determine the aforementioned magic parameters.

[0058] The aforementioned operating conditions can refer to various working conditions or states experienced by vehicle tires in actual use, including but not limited to: longitudinal force conditions, lateral force conditions, and self-centering torque conditions. These conditions can help build a comprehensive understanding of the specific performance of vehicle tires, such as grip, handling stability, and comfort.

[0059] In one alternative embodiment, considering that manually predicting the dynamic characteristics of vehicle tires based on feature vectors may introduce new human errors, resulting in low accuracy of the predicted magic parameters, in order to improve the accuracy of the prediction and thus improve the quality of the subsequent dynamic model construction, the construction system can pre-train a parameter prediction model based on machine learning algorithms. In practical applications, the construction system can input the feature vectors into the model, thereby using the model to predict the dynamic characteristics corresponding to the mechanical behavior of vehicle tires under different force conditions, and thus obtain the dynamic characteristics of vehicle tires under different operating conditions. Subsequently, the construction system can determine more accurate magic parameters based on these dynamic characteristics.

[0060] For example, in order to accurately identify the deep nonlinear mapping relationship between feature vectors and magic formula parameters and significantly improve prediction accuracy, the system can use a deep neural network as a parameter prediction model. This model can effectively capture the complex relationship between tire features and magic formula parameters through multi-layer nonlinear transformations, and can provide accurate prediction results even in nonlinear and high-dimensional spaces.

[0061] For example, to capture the dynamic changes of tires during different operating conditions and improve the prediction accuracy of vehicle tire magic parameters, the system can also use a fusion model of convolutional neural networks and recurrent neural networks as the parameter prediction model. Convolutional neural networks can be used to extract the spatial local correlation of static and dynamic features in the feature vector. Considering that tire performance indicators such as temperature and wear change after prolonged high-speed driving or frequent steering, and these changes affect the tire's magic parameters and thus its mechanical properties, recurrent neural networks can be used to handle the temporal dependence of continuous operating conditions, thereby more accurately capturing the dynamic changes in tire performance and improving the prediction accuracy of magic parameters. The fusion of convolutional neural networks and recurrent neural networks can more comprehensively simulate the dynamic characteristics of vehicle tires under continuous operating conditions.

[0062] In this embodiment, the method involves acquiring feature parameters of a vehicle tire across multiple feature dimensions; encoding the feature parameters based on their parameter types to obtain feature vectors; predicting the dynamic characteristics of the vehicle tire based on these feature vectors to obtain magic parameters; and constructing a dynamic model of the vehicle tire based on these magic parameters. By comprehensively considering the multidimensional features of the vehicle tire, more comprehensive feature parameters can be obtained, providing richer and more accurate data input for magic parameter prediction. Further encoding of the feature parameters converts their format into a digital format suitable for magic parameter prediction, making the prediction of magic parameters more efficient and accurate. This reduces the time required to construct a dynamic model based on magic parameters, achieving the goal of rapidly constructing a dynamic model of the vehicle tire. This improves the efficiency of constructing a dynamic model of the vehicle tire, thus solving the technical problem of low efficiency in constructing dynamic models of vehicle tires in related technologies.

[0063] Furthermore, the parameter prediction model includes a feature extraction layer and multiple parameter prediction layers, with different parameter prediction layers corresponding to different operating conditions. The feature vector is input into the parameter prediction model, which is then used to predict the dynamic characteristics of the vehicle tire under different operating conditions to obtain magic parameters. This includes: using the feature extraction layer to extract features from the feature vector to obtain shared features, where the shared features characterize features used in predicting the dynamic characteristics of the vehicle tire under any operating condition; and using multiple parameter prediction layers to process the shared features to obtain the magic parameters of the vehicle tire under different operating conditions.

[0064] The aforementioned feature extraction layer can be a network layer in the aforementioned parameter prediction model used to extract the features actually used in subsequent prediction of magic parameters from the multidimensional physical features of the tire.

[0065] The aforementioned parameter prediction layer can be the network layer in the aforementioned parameter prediction model used to predict the magic parameters of vehicle tires under different operating conditions.

[0066] The aforementioned shared features can refer to the features extracted by the feature extraction layer from the multi-dimensional physical features of the tire, which are used in the process of predicting the dynamic characteristics under any operating condition.

[0067] In one optional embodiment, considering that the dynamic characteristics of a tire are affected by various features, including but not limited to geometry, internal structure, material properties, and external environmental conditions, and that some of these features may affect the dynamic characteristics of the vehicle tire under all operating conditions, in order to capture the shared features that affect all operating conditions and thus improve the efficiency of subsequent magic parameter prediction, the system can utilize a feature extraction layer to extract features from the feature vector. This allows the system to obtain features used in predicting the dynamic characteristics of the vehicle tire under any operating condition, thereby obtaining the aforementioned shared features. This avoids the repeated acquisition of shared features and improves the prediction efficiency of magic parameters. Furthermore, considering that different physical mechanisms and processing logics may be needed in the prediction of magic parameters under different operating conditions, if the shared features are processed in a single prediction layer, the lack of targeted processing for the current operating condition may result in the final magic parameters not accurately reflecting the current operating condition, thus affecting prediction performance. Therefore, in order to improve the accuracy of prediction, the system can be built with multiple parameter prediction layers, and the shared features can be processed separately by multiple parameter prediction layers to obtain the magic parameters of vehicle tires under different operating conditions. This allows for independent prediction of magic parameters under each different operating condition, avoiding the limitations of a single model in the prediction process and thus improving the accuracy of the prediction results.

[0068] Furthermore, multiple parameter prediction layers are used to process the shared features to obtain the magic parameters of the vehicle tires under different operating conditions. This includes: during the processing of the shared features using any parameter prediction layer, obtaining the degree of correlation between the operating condition corresponding to the parameter prediction layer and any shared feature; based on the degree of correlation, selecting a target feature from the shared features, wherein the degree of correlation corresponding to the target feature is greater than the degree of correlation corresponding to other features in the shared features besides the target feature; and processing the target feature based on the parameter prediction layer to obtain the magic parameters of the vehicle tires under the operating conditions corresponding to the parameter prediction layer.

[0069] The aforementioned target features can be selected from the shared features and found to be highly correlated with the operating conditions corresponding to the parameter prediction layer.

[0070] In one optional embodiment, considering that the relationship between the magic parameters of vehicle tires and shared features is highly non-linear, meaning that the same shared feature may map to different magic parameters under different operating conditions, in order to accurately obtain magic parameters that best match the current operating condition, the system can first determine the degree of correlation between the operating condition corresponding to the parameter prediction layer and any shared feature during the processing of shared features using any parameter prediction layer. Subsequently, based on the degree of correlation, the system can select target features from the shared features whose correlation is greater than that of other features in the shared features, and process these target features based on the parameter prediction layer to obtain the magic parameters of the vehicle tires under the operating condition corresponding to the parameter prediction layer. This allows the parameter prediction layer to automatically identify and focus on features that are highly correlated with the operating condition corresponding to the parameter prediction layer when predicting magic parameters for different operating conditions, thereby reducing the waste of computational resources during the prediction process and improving prediction efficiency and accuracy.

[0071] For example, we can assume three parameter prediction layers, corresponding to the prediction of magic parameters for vehicle tires under three different operating conditions: dry road surface, wet road surface, and icy / snowy road surface. When the system uses the parameter prediction model for prediction, the shared feature extraction layer has already extracted shared features from the multidimensional features of the tire. Next, for the parameter prediction layer for dry road surface, the system can calculate the correlation between this parameter prediction layer and the shared features, thereby determining that the correlation between the vehicle tire's tread pattern features, tread rubber compound, and air pressure and the magic parameters for dry road surface is higher than the correlation between other shared features and the magic parameters for dry road surface. Therefore, the system can select the tread pattern features, tread rubber compound, and air pressure from the aforementioned shared features as the target features, and can process these target features using the parameter prediction layer for dry road surface to obtain the magic parameters of the vehicle tire under dry road surface operating conditions. The same prediction process also applies to the parameter prediction layers for wet road surface and icy / snowy road surface.

[0072] The number of parameter prediction layers and application scenarios mentioned above are only illustrative examples. Staff can set them according to their actual needs, and there are no restrictions here.

[0073] Furthermore, the operating conditions include: longitudinal force condition, lateral force condition, and self-centering torque condition. The longitudinal force condition characterizes the mechanical behavior of the vehicle tires during straight-line driving; the lateral force condition characterizes the mechanical behavior of the vehicle tires during steering; and the self-centering torque condition characterizes the mechanical behavior of the vehicle tires during the process of resuming straight-line driving after steering. Multiple parameter prediction layers include: a longitudinal force parameter prediction layer, a lateral force parameter prediction layer, and a self-centering torque prediction layer. Based on the parameter prediction layers, the target features are processed to obtain... The magic parameters for the vehicle tires under the corresponding operating conditions in the parameter prediction layer include: processing the target features based on the longitudinal force parameter prediction layer to obtain the longitudinal force magic parameters of the vehicle tires under the longitudinal force condition; processing the target features based on the lateral force parameter prediction layer to obtain the lateral force magic parameters of the vehicle tires under the lateral force condition; processing the target features based on the self-aligning torque parameter prediction layer to obtain the self-aligning torque magic parameters of the vehicle tires under the self-aligning torque condition; and obtaining the magic parameters based on the longitudinal force magic parameters, lateral force magic parameters, and self-aligning torque magic parameters.

[0074] The aforementioned longitudinal force conditions refer to the main force conditions along the vehicle's direction of travel generated when the vehicle's tires contact the ground during linear acceleration or deceleration. Under longitudinal force conditions, the vehicle's tires mainly bear longitudinal friction forces, including braking and driving forces, which affect the tires' grip and the vehicle's acceleration and deceleration performance.

[0075] The aforementioned lateral force conditions can be the main lateral force generated when the vehicle tires contact the ground during steering or driving in crosswind conditions. Under lateral force conditions, the vehicle tires are affected by lateral force, which determines the vehicle's handling stability and cornering ability.

[0076] The aforementioned self-centering torque condition refers to the torque condition generated by the natural self-centering tendency of the tire structure and material properties when the vehicle tires tend to return to a straight-line driving state after the vehicle has finished turning. This helps to automatically correct the vehicle's direction and maintain the stability of straight-line driving.

[0077] The aforementioned longitudinal force parameter prediction layer can be a network layer used to predict magic parameters under longitudinal force conditions from multidimensional features of vehicle tires.

[0078] The aforementioned lateral force parameter prediction layer can be a network layer used to predict magic parameters under lateral force conditions from multidimensional features of vehicle tires.

[0079] The aforementioned self-aligning torque prediction layer can be a network layer used to predict the magic parameters of vehicle tires under self-aligning torque conditions.

[0080] The aforementioned longitudinal force magic parameters can be magic parameters used to describe the mechanical behavior of vehicle tires under longitudinal force conditions. These parameters may include, but are not limited to, the tire's longitudinal stiffness coefficient, friction coefficient, slip ratio threshold, etc., which determine the tire's response characteristics under different longitudinal force conditions.

[0081] The aforementioned lateral force magic parameters can be magic parameters used to describe the mechanical behavior of vehicle tires under lateral force conditions. They may include, but are not limited to, the lateral stiffness and slip angle sensitivity of vehicle tires, reflecting the deformation and force transmission characteristics of vehicle tires under lateral force.

[0082] The aforementioned magic parameters of the return torque can be parameters used to describe the characteristics of the tire return torque under working conditions. They may include, but are not limited to, the natural return coefficient of the vehicle tire and the contribution of the tire sidewall elasticity to the return torque. They are important bases for analyzing the vehicle's steering recovery ability and stability.

[0083] In one optional embodiment, considering that longitudinal force, lateral force, and self-aligning torque reflect the mechanical behavior of vehicle tires under different dynamic conditions, and that the dynamic characteristics exhibited by different mechanical behaviors are significantly different, in order to more accurately determine the dynamic characteristics of vehicle tires and thus improve the accuracy of subsequent dynamic model construction, the construction system can set up three parameter prediction layers: a longitudinal force parameter prediction layer, a lateral force parameter prediction layer, and a self-aligning torque parameter prediction layer. The longitudinal force parameter prediction layer is used to predict the magic parameter corresponding to the longitudinal force condition, the lateral force parameter prediction layer is used to predict the magic parameter corresponding to the lateral force condition, and the self-aligning torque parameter prediction layer is used to predict the magic parameter corresponding to the self-aligning torque condition. This approach can avoid mutual interference in the prediction process under different conditions, thereby improving the prediction accuracy of the magic parameter.

[0084] Specifically, longitudinal force is closely related to tire grip, friction coefficient, and tire compression and tension behavior. The system can process target features based on the longitudinal force parameter prediction layer to obtain the longitudinal force magic parameters of the vehicle tire under longitudinal force conditions, so as to accurately reflect the longitudinal dynamic characteristics of the tire.

[0085] Lateral force is generated when a vehicle is turning or under lateral acceleration. It involves the tire's lateral behavior and lateral stiffness. In order to predict the magic parameters under lateral force conditions from the structure, material and geometric features of the tire, the system can process the target features based on the lateral force parameter prediction layer to obtain the lateral force magic parameters of the vehicle tire under lateral force conditions.

[0086] Furthermore, when a tire deflects, a torque is generated within a certain angle to bring the tire back to a straight-line driving state, which is called the return torque. The return torque is related to the tire's sidewall stiffness and tread shape, and is the key to maintaining the vehicle's straight-line stability and automatic return capability. In order to accurately obtain the magic parameters corresponding to the return torque condition, the system can process the target features based on the return torque parameter prediction layer to obtain the vehicle tire's return torque magic parameters under the return torque condition.

[0087] Ultimately, the system can integrate magic parameters under longitudinal force conditions, lateral force conditions, and restoring moment conditions to form a complete set of magic parameters, thereby improving the accuracy of subsequent dynamic model construction.

[0088] For ease of understanding, Figure 5 This is a schematic diagram of an optional parameter prediction model according to an embodiment of this application, such as... Figure 5 As shown, the model consists of an input layer, a shared feature extraction layer, a task-specific prediction layer, and an output layer. The input layer receives multi-dimensional tire features. The shared feature extraction layer comprises three fully connected sub-layers (not shown in the figure), each using the Rectified Linear Unit (ReLU) activation function, with 64 neurons per layer. This allows for the gradual abstraction and extraction of high-level representations of tire features through deep learning. Specifically, the first fully connected sub-layer captures basic feature combinations, the second layer learns intermediate semantic features, and the third layer constructs high-level abstract features. These features are crucial for understanding and predicting tire behavior, and the ReLU activation function ensures the model can handle non-linear relationships between features, improving its expressive and generalization capabilities. The task-specific prediction layer (i.e., the aforementioned parameter prediction layer) is placed after the shared feature extraction layer. The model employs three independent task-specific prediction layers (i.e., the longitudinal force parameter prediction layer, the lateral force parameter prediction layer, and the self-aligning torque prediction layer), corresponding to longitudinal force parameter prediction, lateral force parameter prediction, and self-aligning torque parameter prediction, respectively. Each prediction branch is a two-layer fully connected network. The first layer has 32 neurons, and the number of neurons in the second layer is determined by the number of magic parameters to be predicted. This architectural design allows the model to learn and predict magic parameters specifically for each operating condition. Furthermore, through shared feature representations, it effectively promotes information sharing and generalization between tasks. The model's output layer is directly connected to the task-specific layers and outputs the predicted magic formula parameters. Specifically, the output layer provides sets of magic parameters for longitudinal force, lateral force, and self-aligning torque operating conditions. These parameters can be directly used for vehicle dynamics simulation and control system design, eliminating the need for expensive and time-consuming bench tests.

[0089] The specific values ​​for fully connected sublayers, number of neurons, etc. mentioned above are for illustrative purposes only. Staff can set them according to their actual needs, and there are no restrictions here.

[0090] Figure 6 This is a schematic diagram of the training process of an optional parameter prediction model according to an embodiment of this application, as shown below. Figure 6 As shown, firstly, the system can divide the dataset based on the tire feature-parameter mapping database. For example, the dataset can be divided into an 80% training set, a 10% test set, and a 10% validation set. Then, the system can build a model based on PyTorch (an open-source machine learning library), train the model using the training set, test the model using the test set, and validate the model's prediction performance using the validation set, thus obtaining the final parameter prediction model based on multi-dimensional features.

[0091] The above dataset division is for illustrative purposes only. Staff can set it according to their actual needs, and there are no restrictions here.

[0092] Furthermore, the feature parameters are encoded based on their parameter types to obtain feature vectors, including: obtaining the parameter padding method and parameter encoding method corresponding to the feature parameters based on their parameter types; padding the feature parameters based on the parameter padding method to obtain padding parameters; performing outlier detection on the padding parameters to obtain outlier detection results; and encoding the padding parameters using the parameter encoding method when the outlier detection results indicate that there are no outliers in the padding parameters to obtain feature vectors.

[0093] The parameter imputation methods described above refer to methods for handling missing values ​​in feature parameters. These methods vary depending on the parameter type. For example, for numerical parameters, the above imputation method can employ K-nearest neighbor-based interpolation, which fills missing values ​​based on the average or weighted average of the K nearest neighbors in the feature parameter space. As another example, for categorical parameters, the above imputation method can use mode imputation, that is, filling missing values ​​with values ​​that occur most frequently in that category.

[0094] The above parameter encoding method can refer to the way of converting feature parameters into a format readable by machine learning models. For example, for classification parameters, the above parameter encoding method can adopt one-hot encoding, which can convert each classification parameter into a binary vector, with only the corresponding classification position being 1 and the rest being 0, which makes it easier for the model to process classification information.

[0095] The aforementioned padding parameters can be feature parameters processed through parameter padding. These parameters appropriately fill in missing values, making the dataset complete and usable for subsequent data preprocessing and model training.

[0096] The above outlier detection results can be obtained by analyzing the filling parameters through an outlier detection algorithm to determine whether there are outliers that exceed the normal range. If outliers are found, further processing or removal may be required. If no outliers are found, the parameters can be considered valid and can be used for subsequent coding and modeling.

[0097] In one optional embodiment, considering that feature parameters may contain missing values, the construction system needs to impute these missing values ​​to ensure the accurate construction of the feature vector. Different types of feature parameters may correspond to different missing value imputation strategies. Therefore, to effectively impute missing values ​​for different types of feature parameters, the construction system can first obtain the parameter imputation method corresponding to the feature parameter based on the parameter type, thereby ensuring the integrity and consistency of the data and avoiding bias or misinterpretation due to missing values, which could lead to low accuracy in the final constructed dynamic model. Furthermore, considering that some non-numerical feature parameters are difficult for the parameter prediction model to understand directly, to ensure the prediction accuracy of the magic parameter, these feature parameters need to be converted into numerical formats first. Therefore, the construction system can also improve the parameter prediction model's understanding of classification parameters based on the parameter type, thereby improving the overall prediction performance of the parameter prediction model. Subsequently, the construction system can impute the feature parameters based on the parameter imputation method to obtain the imputed parameters, thus supplementing the missing values. After obtaining the aforementioned imputation parameters, the system can perform outlier detection to determine if any outliers exist, thus obtaining outlier detection results. This allows for accurate identification and removal of outliers, further purifying the dataset and improving the overall quality of the imputation parameters. Finally, if the outlier detection results indicate that no outliers exist in the imputation parameters, the system can encode the imputation parameters using a parameter encoding method to obtain feature vectors. This allows the encoded feature parameters to be presented in numerical form, facilitating understanding and calculation by the parameter prediction model.

[0098] Furthermore, the parameter types include: numerical parameters and categorical parameters. Numerical parameters are used to characterize that the feature parameters are numerically meaningful, while categorical parameters are used to characterize that the feature parameters are physical properties of vehicle tires. The feature parameters are filled based on the parameter filling method to obtain filled parameters, including: when the parameter type is numerical, interpolating the feature parameters to obtain filled parameters; when the parameter type is categorical, performing mode filling on the feature parameters to obtain filled parameters.

[0099] The above numerical parameters can be data with numerical meaning, including but not limited to: tire width, diameter, etc.

[0100] The above classification parameters can be data with category significance, including but not limited to: tire tread type, temperature rating, etc.

[0101] In an optional embodiment, considering that different types of feature parameters require different imputation strategies to ensure that the imputation of missing values ​​is effective and accurate, the system can classify the above-mentioned parameter types into numerical parameters and categorical parameters to facilitate the adoption of targeted imputation strategies. Specifically, when the parameter type is numerical, the system can interpolate the feature parameters to obtain the imputed parameters. When the parameter type is categorical, the system can perform mode imputation on the feature parameters to obtain the imputed parameters. Through interpolation and mode imputation, the obtained imputed parameters can effectively reflect the true distribution of vehicle tire features, providing an accurate data foundation for subsequent magic parameter prediction.

[0102] For example, to improve the accuracy of missing value imputation, for numerical parameters such as tire cross-section width and tire aspect ratio, the system can use the K-nearest neighbor interpolation method to imput missing values ​​and obtain the imputed parameters. For categorical parameters such as temperature rating and wear index, the system can use mode imputation to obtain the imputed parameters.

[0103] Furthermore, the filler parameters are encoded using a parametric encoding method to obtain the feature vector, including: when the parameter type is a classification parameter, the feature parameters are one-hot encoded to obtain the feature vector.

[0104] In one alternative embodiment, considering that classification parameters are difficult to process directly by the parameter prediction model, the system needs to first convert the classification parameters into feature vectors through encoding. Based on this, the system can perform one-hot encoding on the feature parameters when the parameter type is classification parameter to obtain feature vectors. This encoding strategy can effectively convert the tire classification parameters into numerical data, enabling the parameter prediction model to learn the force and torque response characteristics of different types of tires under different working conditions, thereby improving the accuracy of magic parameter prediction.

[0105] For ease of understanding, Figure 7 This is a schematic diagram of an optional parameter processing procedure according to an embodiment of this application, such as... Figure 7As shown, the system can process the parameters in the tire feature-parameter mapping database according to the parameter type. Specifically, for categorical parameters, the system can perform feature encoding to obtain temperature rating, wear index, and traction rating. Subsequently, the system can perform one-heat encoding on these parameters to convert them into numerical form so that the parameter prediction model can understand and process them.

[0106] For numerical parameters of tires, such as tire cross-sectional width, tire aspect ratio, and rim diameter, the system can use a K-nearest neighbor interpolation algorithm to fill in missing values. This algorithm estimates missing values ​​based on the nearest data points. For categorical parameters, such as temperature rating, wear index, and traction rating, the system can use the mode to fill in missing values, thereby ensuring that all sample data are complete and avoiding missing values ​​from affecting subsequent analysis and parameter prediction model training.

[0107] In addition, to identify and handle any outlier data points that may exist in the filling parameters, the system can use the isolated forest algorithm to detect outlier data points in the filling parameters, so as to ensure that there are no extreme values ​​or noise points in the filling parameters, thereby improving the purity of the filling parameters and the prediction accuracy of the parameter prediction model.

[0108] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of the relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0109] According to an embodiment of this application, a device for constructing a dynamic model is provided. It should be noted that this device can be used to execute the aforementioned method for constructing a dynamic model. The specific implementation methods and application scenarios are the same as those in the above embodiments, and will not be repeated here. Figure 8 This is a schematic diagram of a dynamic model construction apparatus according to an embodiment of this application, such as... Figure 8 As shown, the device includes:

[0110] The first acquisition module 802 is used to acquire the feature parameters of the vehicle tires in multiple feature dimensions.

[0111] The first encoding module 804 is used to encode the feature parameters based on the parameter type of the feature parameters to obtain the feature vector of the feature parameters.

[0112] The first prediction module 806 is used to predict the dynamic characteristics of vehicle tires based on feature vectors to obtain the magic parameters of vehicle tires.

[0113] The first building module 808 is used to build a dynamic model of the vehicle tires based on magic parameters.

[0114] Furthermore, the first prediction module is also used to: input the feature vector into the parameter prediction model, use the parameter prediction model to predict the dynamic characteristics of the vehicle tire under different operating conditions, and obtain magic parameters, wherein the operating conditions are used to characterize the mechanical behavior of the vehicle tire under different force conditions.

[0115] Furthermore, the parameter prediction model includes a feature extraction layer and multiple parameter prediction layers, with different parameter prediction layers corresponding to different operating conditions. The first prediction module is also used to: extract features from the feature vector using the feature extraction layer to obtain shared features, wherein the shared features are used to characterize the features used in the process of predicting the dynamic characteristics of the vehicle tire under any operating condition; and process the shared features using multiple parameter prediction layers to obtain the magic parameters of the vehicle tire under different operating conditions.

[0116] Furthermore, the first prediction module is also used to: obtain the degree of correlation between the operating condition corresponding to the parameter prediction layer and any shared feature during the process of processing the shared feature using any parameter prediction layer; select a target feature from the shared features based on the degree of correlation, wherein the degree of correlation corresponding to the target feature is greater than the degree of correlation corresponding to other features in the shared features excluding the target feature; and process the target feature based on the parameter prediction layer to obtain the magic parameters of the vehicle tire under the operating condition corresponding to the parameter prediction layer.

[0117] Furthermore, the operating conditions include: longitudinal force condition, lateral force condition, and self-aligning torque condition. The longitudinal force condition characterizes the mechanical behavior of the vehicle tires during straight-line driving; the lateral force condition characterizes the mechanical behavior of the vehicle tires during steering; and the self-aligning torque condition characterizes the mechanical behavior of the vehicle tires during straight-line driving after steering. Multiple parameter prediction layers include: a longitudinal force parameter prediction layer, a lateral force parameter prediction layer, and a self-aligning torque prediction layer. The first prediction module is also used to: process target features based on the longitudinal force parameter prediction layer to obtain the longitudinal force magic parameter of the vehicle tires under the longitudinal force condition; process target features based on the lateral force parameter prediction layer to obtain the lateral force magic parameter of the vehicle tires under the lateral force condition; process target features based on the self-aligning torque parameter prediction layer to obtain the self-aligning torque magic parameter of the vehicle tires under the self-aligning torque condition; and obtain the magic parameter based on the longitudinal force magic parameter, lateral force magic parameter, and self-aligning torque magic parameter.

[0118] Furthermore, the first encoding module is also used to: obtain the parameter filling method and parameter encoding method corresponding to the feature parameter based on the parameter type; fill the feature parameter based on the parameter filling method to obtain the filled parameter; perform outlier detection on the filled parameter to obtain the outlier detection result; and encode the filled parameter using the parameter encoding method when the outlier detection result indicates that there is no outlier in the filled parameter to obtain the feature vector.

[0119] Furthermore, the parameter types include: numerical parameters and categorical parameters. Numerical parameters are used to characterize that the feature parameters are parameters with numerical meaning, while categorical parameters are used to characterize that the feature parameters are parameters describing the physical properties of vehicle tires. The first encoding module is also used to: when the parameter type is a numerical parameter, perform interpolation processing on the feature parameters to obtain filler parameters; when the parameter type is a categorical parameter, perform mode filling on the feature parameters to obtain filler parameters.

[0120] Furthermore, the first encoding module is also used to: perform one-hot encoding on the feature parameters to obtain the feature vector when the parameter type is a classification parameter.

[0121] Embodiments of this application also provide a vehicle, including: a memory storing an executable program; and a processor for running the program, wherein the program executes the methods described in various embodiments of this application when it runs.

[0122] Embodiments of this application also provide a computer-readable storage medium including a stored executable program, wherein, when the executable program is running, it controls the device where the computer-readable storage medium is located to perform the methods of various embodiments of this application.

[0123] Embodiments of this application also provide a computer program product, including a computer program that, when executed by a processor, implements the methods of various embodiments of this application.

[0124] Embodiments of this application also provide a computer program product, including a non-volatile computer-readable storage medium for storing a computer program that, when executed by a processor, implements the methods in various embodiments of this application.

[0125] Embodiments of this application also provide a computer program that, when executed by a processor, implements the methods described in the various embodiments of this application.

[0126] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0127] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0128] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0129] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0130] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard drive, magnetic disk, or optical disk.

[0131] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for constructing a dynamic model, characterized in that, include: Obtain the feature parameters of the vehicle tires across multiple feature dimensions; The feature parameters are encoded based on their parameter types to obtain feature vectors. Based on the feature vector, the dynamic characteristics of the vehicle tire are predicted to obtain the magic parameters of the vehicle tire. Based on the magic parameters, a dynamic model of the vehicle tires is constructed.

2. The method according to claim 1, characterized in that, Based on the feature vector, the dynamic characteristics of the vehicle tire are predicted to obtain the magic parameters of the vehicle tire, including: The feature vector is input into the parameter prediction model, and the parameter prediction model is used to predict the dynamic characteristics of the vehicle tire under different operating conditions to obtain the magic parameters. The operating conditions are used to characterize the mechanical behavior of the vehicle tire under different stress conditions.

3. The method according to claim 2, characterized in that, The parameter prediction model includes a feature extraction layer and multiple parameter prediction layers, with different parameter prediction layers corresponding to different operating conditions. The feature vector is input into the parameter prediction model, which is then used to predict the dynamic characteristics of the vehicle tires under different operating conditions to obtain the magic parameters, including: The feature extraction layer is used to extract features from the feature vector to obtain shared features, wherein the shared features are used to characterize the features used in the process of predicting the dynamic characteristics of the vehicle tires under any operating condition; The shared features are processed by the multiple parameter prediction layers to obtain the magic parameters of the vehicle tires under different operating conditions.

4. The method according to claim 3, characterized in that, The shared features are processed using the multiple parameter prediction layers to obtain the magic parameters of the vehicle tires under different operating conditions, including: During the process of processing the shared feature using any parameter prediction layer, the degree of correlation between the operating condition of the parameter prediction layer and any shared feature is obtained; Based on the degree of association, a target feature is selected from the shared features, wherein the degree of association corresponding to the target feature is greater than the degree of association corresponding to other features in the shared features excluding the target feature; Based on the parameter prediction layer, the target features are processed to obtain the magic parameters of the vehicle tires under the operating conditions corresponding to the parameter prediction layer.

5. The method according to claim 4, characterized in that, The operating conditions include: longitudinal force condition, lateral force condition, and self-centering torque condition. The longitudinal force condition characterizes the mechanical behavior of the vehicle tires during straight-line driving. The lateral force condition characterizes the mechanical behavior of the vehicle tires during steering. The self-centering torque condition characterizes the mechanical behavior of the vehicle tires during straight-line driving after steering. The multiple parameter prediction layers include: a longitudinal force parameter prediction layer, a lateral force parameter prediction layer, and a self-centering torque prediction layer. Based on the parameter prediction layers, the target features are processed to obtain the magic parameters of the vehicle tires under the operating conditions corresponding to the parameter prediction layers, including: Based on the longitudinal force parameter prediction layer, the target features are processed to obtain the longitudinal force magic parameters of the vehicle tire under the longitudinal force condition. The target features are processed based on the lateral force parameter prediction layer to obtain the lateral force magic parameters of the vehicle tire under the lateral force condition. The target features are processed based on the self-aligning torque parameter prediction layer to obtain the self-aligning torque magic parameter of the vehicle tire under the self-aligning torque condition. The magic parameters are obtained based on the longitudinal force magic parameters, the lateral force magic parameters, and the aligning torque magic parameters.

6. The method according to any one of claims 1-5, characterized in that, The feature parameters are encoded based on their parameter types to obtain a feature vector, including: Based on the parameter type, obtain the parameter filling method and parameter encoding method corresponding to the feature parameter; The feature parameters are filled based on the parameter filling method to obtain the filled parameters; Anomaly detection is performed on the filling parameters to obtain the anomaly detection results; If the outlier detection result indicates that there are no outliers in the padding parameters, the padding parameters are encoded using the parameter encoding method to obtain the feature vector.

7. The method according to claim 6, characterized in that, The parameter types include: numerical parameters and categorical parameters. Numerical parameters characterize the feature parameter as having numerical meaning, while categorical parameters characterize the feature parameter as describing the physical properties of the vehicle tire. The feature parameter is filled based on the parameter filling method to obtain filled parameters, including: When the parameter type is the numerical parameter, the feature parameter is interpolated to obtain the filling parameter; When the parameter type is the classification parameter, the feature parameter is padded with the mode to obtain the padded parameter.

8. The method according to claim 7, characterized in that, The feature vector is obtained by encoding the padding parameters using the aforementioned parameter encoding method, including: When the parameter type is the classification parameter, the feature parameter is one-hot encoded to obtain the feature vector.

9. A vehicle, characterized in that, include: Memory, which stores executable programs; A processor for running the program, wherein the program, when running, performs the method according to any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored executable program, wherein, when the executable program is executed, it controls the device on which the storage medium is located to perform the method according to any one of claims 1 to 8.

Images