Construction method, device and equipment of vehicle simulation test platform, and storage medium

Abstract

The invention provides a construction method and device of a vehicle simulation test platform, equipment and a storage medium. The method comprises the following steps: acquiring road surface image data of a vehicle running in at least one road type, speed information of the vehicle in the at least one road type, and curb mass of the vehicle; inputting the at least one road surface image data into a preset convolutional neural network model to obtain target road surface condition information corresponding to the at least one road type; and determining a tire model in the vehicle simulation test platform based on the slip rate of each tire in the vehicle corresponding to the at least one piece of speed information, the at least one piece of target pavement condition information and the curb mass of the vehicle. The method is used for achieving the purpose of improving the test capability and accuracy of the vehicle simulation platform.

Description

Technical Field

[0001] The present application relates to the field of vehicle technology, and in particular to a method, device, equipment, and storage medium for constructing a vehicle simulation test platform. Background Art

[0002] With the gradual deepening of intelligent automotive control systems, the demand for vehicle controllers to control vehicles and their tires is becoming increasingly strong. In order to achieve the testing and verification of high-level vehicle controller control strategies, it is of great significance to develop vehicle software functions by building a simulation test platform to perform model-in-the-loop or hardware-in-the-loop control strategies.

[0003] In the existing technology, the model method is often used to construct a vehicle simulation test platform. However, in actual application, due to the large fluctuations in tire slip rate during actual driving of the vehicle, it is necessary to use algorithms such as second-order filtering or Kalman filtering to filter the collected vehicle body status signals. This has a poor effect on real-time vehicle control and cannot obtain accurate parameters to build an accurate and high-performance vehicle simulation test platform.

[0004] Based on the above existing technologies, how to construct a vehicle simulation test platform with better testing capabilities and accuracy has become a technical problem that needs to be solved urgently. Summary of the Invention

[0005] The embodiments of the present application provide a method, apparatus, equipment, and storage medium for constructing a vehicle simulation test platform to achieve the purpose of improving the testing capability and accuracy of the vehicle simulation platform.

[0006] In a first aspect, an embodiment of the present application provides a method for constructing a vehicle simulation test platform, comprising:

[0007] Acquiring road surface image data of a vehicle traveling on at least one road type, speed information of the vehicle on the at least one road type, and curb weight of the vehicle;

[0008] Inputting the at least one road surface image data into a preset convolutional neural network model to obtain target road surface condition information corresponding to the at least one road type;

[0009] A tire model in the vehicle simulation test platform is determined based on the slip rate of each tire in the vehicle corresponding to the at least one speed information, the at least one target road condition information, and the curb weight of the vehicle.

[0010] In one or more embodiments, inputting the at least one road surface image data into a preset convolutional neural network model to obtain target road surface condition information corresponding to the at least one road type includes:

[0011] For each road surface image data, sequentially performing convolution processing, maximum pooling, and data flattening processing on the road surface image data to obtain a fully connected layer corresponding to at least one single-dimensional data group corresponding to the road surface image data, wherein the convolution processing is performed in a preset number of image processing channels in the convolutional neural network model, the preset number being obtained based on RGB values, and for each single-dimensional data, the single-dimensional data is used to mark the abstract features of the pixels in the road surface image data;

[0012] Obtaining target data results based on mapping and computational processing between the at least one fully connected layer and the neural network node input and hidden layer in the convolutional neural network model;

[0013] The target road condition information corresponding to the road image data is obtained by matching the target data result with at least one preset road condition information.

[0014] In one or more embodiments, the step of sequentially performing convolution processing, maximum pooling, and data flattening processing on the road surface image data to obtain a fully connected layer corresponding to at least one single-dimensional data group corresponding to the road surface image data includes:

[0015] Performing the convolution process on the road surface image data to obtain a first feature map set corresponding to the road surface image data;

[0016] performing the maximum pooling and the data flattening processing on the first feature map set in sequence to form a first fully connected layer corresponding to the first single-dimensional data group;

[0017] performing the maximum pooling and the data flattening processing on the second feature map set in sequence to form a second fully connected layer corresponding to the second single-dimensional data group, wherein the second feature map set is obtained by performing the maximum pooling processing and then the convolution processing on the first feature map set;

[0018] performing the maximum pooling and the data flattening processing on a third feature map set in sequence to form a third fully connected layer corresponding to a third one-dimensional data group, wherein the third feature map set is obtained by performing the maximum pooling processing and then the convolution processing on the second feature map set;

[0019] The fully connected layers corresponding to the at least one single-dimensional data group include: the first fully connected layer, the second fully connected layer, and the third fully connected layer.

[0020] In one or more embodiments, for each piece of speed information, the speed information includes: wheel speed information of at least one wheel and acceleration information of the vehicle;

[0021] Accordingly, before determining the tire model in the vehicle simulation test platform based on the slip rate of each tire in the vehicle corresponding to the at least one speed information, the at least one target road condition information, and the curb weight of the vehicle, the method further includes:

[0022] Using an extended Kalman filter algorithm, fusing the wheel speed information of the at least one wheel and the acceleration information of the vehicle to obtain actual speed information of the vehicle;

[0023] For each wheel, a slip rate of a tire corresponding to the wheel is determined based on the actual speed information and the wheel speed information of the wheel.

[0024] In one or more embodiments, the adopting of the extended Kalman filter algorithm to fuse the wheel speed information of the at least one wheel and the acceleration information of the vehicle to obtain the actual speed information of the vehicle includes:

[0025] Determining a state equation and an observation equation of an observer in the extended Kalman filter algorithm based on the wheel speed information of the at least one wheel and the acceleration information of the vehicle, the state equation comprising: a yaw angular acceleration equation, a center of mass sideslip angular velocity equation, and a longitudinal acceleration equation of the vehicle, and the observation equation comprising: a center of mass lateral acceleration equation;

[0026] The actual speed information of the vehicle is determined according to the general form equation of the observer in the extended Kalman filter algorithm, the state equation, and the observation equation.

[0027] In one or more embodiments, the target road surface condition information includes: a peak road surface adhesion coefficient;

[0028] Accordingly, determining a tire model in the vehicle simulation test platform based on the slip rate of each tire in the vehicle corresponding to the at least one speed information, the at least one target road condition information, and the curb weight of the vehicle includes:

[0029] Determining a calculation formula for the actual maximum grip of each tire corresponding to the at least one road type based on the peak road surface adhesion coefficient corresponding to the at least one road type, the slip rate of each tire corresponding to the at least one road type, and the curb weight of the vehicle;

[0030] The tire model in the vehicle simulation test platform is determined according to a calculation formula for the actual maximum grip of each tire corresponding to the at least one road type.

[0031] In one or more embodiments, the calculation formula for determining the actual maximum grip of each tire corresponding to the at least one road type based on the peak road surface adhesion coefficient corresponding to the at least one road type, the slip rate of each tire corresponding to the at least one road type, and the curb weight of the vehicle includes:

[0032] For each tire, determining a theoretical maximum grip of the tire in longitudinal travel based on a maximum peak adhesion coefficient among the road surface peak adhesion coefficients and a vertical pressure exerted by the tire on the ground;

[0033] A calculation formula for the actual maximum grip of the tire corresponding to the at least one road type is determined based on the theoretical maximum grip, the slip rate of the tire corresponding to the at least one road type, and the curb weight of the vehicle.

[0034] In a second aspect, an embodiment of the present application provides a device for constructing a vehicle simulation test platform, comprising:

[0035] an acquisition module, configured to acquire road surface image data of a vehicle traveling on at least one road type, speed information of the vehicle on the at least one road type, and curb weight of the vehicle;

[0036] a processing module, configured to input the at least one road surface image data into a preset convolutional neural network model to obtain target road surface condition information corresponding to the at least one road type;

[0037] A determination module is used to determine a tire model in the vehicle simulation test platform based on the slip rate of each tire in the vehicle corresponding to the at least one speed information, the at least one target road condition information, and the curb weight of the vehicle.

[0038] In one or more embodiments, the processing module is specifically configured to:

[0039] For each road surface image data, sequentially performing convolution processing, maximum pooling, and data flattening processing on the road surface image data to obtain a fully connected layer corresponding to at least one single-dimensional data group corresponding to the road surface image data, wherein the convolution processing is performed in a preset number of image processing channels in the convolutional neural network model, the preset number being obtained based on RGB values, and for each single-dimensional data, the single-dimensional data is used to mark the abstract features of the pixels in the road surface image data;

[0040] Obtaining target data results based on mapping and computational processing between the at least one fully connected layer and the neural network node input and hidden layer in the convolutional neural network model;

[0041] The target road condition information corresponding to the road image data is obtained by matching the target data result with at least one preset road condition information.

[0042] In one or more embodiments, the processing module sequentially performs convolution processing, maximum pooling, and data flattening processing on the road image data to obtain a fully connected layer corresponding to at least one single-dimensional data group corresponding to the road image data, specifically for:

[0043] Performing the convolution process on the road surface image data to obtain a first feature map set corresponding to the road surface image data;

[0044] performing the maximum pooling and the data flattening processing on the first feature map set in sequence to form a first fully connected layer corresponding to the first single-dimensional data group;

[0045] performing the maximum pooling and the data flattening processing on the second feature map set in sequence to form a second fully connected layer corresponding to the second single-dimensional data group, wherein the second feature map set is obtained by performing the maximum pooling processing and then the convolution processing on the first feature map set;

[0046] performing the maximum pooling and the data flattening processing on a third feature map set in sequence to form a third fully connected layer corresponding to a third one-dimensional data group, wherein the third feature map set is obtained by performing the maximum pooling processing and then the convolution processing on the second feature map set;

[0047] The fully connected layers corresponding to the at least one single-dimensional data group include: the first fully connected layer, the second fully connected layer, and the third fully connected layer.

[0048] In one or more embodiments, for each piece of speed information, the speed information includes: wheel speed information of at least one wheel and acceleration information of the vehicle;

[0049] Accordingly, before determining the tire model in the vehicle simulation test platform based on the slip rate of each tire in the vehicle corresponding to the at least one speed information, the at least one target road condition information, and the curb weight of the vehicle, the processing module is further configured to:

[0050] Using an extended Kalman filter algorithm, fusing the wheel speed information of the at least one wheel and the acceleration information of the vehicle to obtain actual speed information of the vehicle;

[0051] For each wheel, a slip rate of a tire corresponding to the wheel is determined based on the actual speed information and the wheel speed information of the wheel.

[0052] In one or more embodiments, the processing module fuses the wheel speed information of the at least one wheel and the acceleration information of the vehicle by using an extended Kalman filtering algorithm to obtain actual speed information of the vehicle, and is specifically configured to:

[0053] determine a state equation and an observation equation of an observer in the extended Kalman filtering algorithm according to the wheel speed information of the at least one wheel and the acceleration information of the vehicle, wherein the state equation comprises a yaw angular acceleration equation, a mass center side slip angular velocity equation and a longitudinal acceleration equation of the vehicle, and the observation equation comprises a mass center lateral acceleration equation;

[0054] determine the actual speed information of the vehicle according to a general form equation of the observer, the state equation and the observation equation in the extended Kalman filtering algorithm.

[0055] In one or more embodiments, the target road surface condition information comprises a peak value adhesion coefficient of a road surface;

[0056] Correspondingly, the determining module is specifically configured to:

[0057] determine a calculation formula of actual maximum grip force of each tire corresponding to the at least one road type according to the peak value adhesion coefficient of the road surface corresponding to the at least one road type, the slip ratio of each tire corresponding to the at least one road type, and the kerb mass of the vehicle;

[0058] determine a tire model in the vehicle simulation test platform according to the calculation formula of the actual maximum grip force of each tire corresponding to the at least one road type.

[0059] In one or more embodiments, the determining module determines the calculation formula of the actual maximum grip force of each tire corresponding to the at least one road type according to the peak value adhesion coefficient of the road surface corresponding to the at least one road type, the slip ratio of each tire corresponding to the at least one road type, and the kerb mass of the vehicle, and is specifically configured to:

[0060] for each tire, determine a theoretical maximum grip force of the tire in the longitudinal direction according to a maximum peak value adhesion coefficient in the peak value adhesion coefficient and a vertical pressure of the tire on the ground;

[0061] determine the calculation formula of the actual maximum grip force of the tire corresponding to the at least one road type according to the theoretical maximum grip force, the slip ratio of the tire corresponding to the at least one road type, and the kerb mass of the vehicle.

[0062] In a third aspect, an electronic device is provided, comprising a memory and a processor.

[0063] The memory stores computer-executable instructions;

[0064] The processor executes the computer-executable instructions stored in the memory, so that the processor executes the above first aspect and / or various possible implementations of the first aspect.

[0065] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, they are used to implement the first aspect above and / or various possible implementation methods of the first aspect.

[0066] In a fifth aspect, an embodiment of the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the above first aspect and / or various possible implementation methods of the first aspect.

[0067] The method, apparatus, equipment, and storage medium for constructing a vehicle simulation test platform provided in an embodiment of the present application obtain road surface image data of a vehicle traveling on at least one road type, as well as speed information of the vehicle on at least one road type, and the curb weight of the vehicle; input the at least one road surface image data into a preset convolutional neural network model to obtain target road surface condition information corresponding to at least one road type; and determine the tire model in the vehicle simulation test platform based on the slip rate of each tire in the vehicle corresponding to the at least one speed information, at least one target road surface condition information, and the curb weight of the vehicle. In this technical solution, by obtaining road surface image data, speed information, and curb weight of a vehicle traveling on at least one road type, multi-dimensional vehicle driving-related data is collected, and the road surface image data is input into a preset convolutional neural network model. Utilizing the model's powerful image recognition and feature extraction capabilities, the target road surface condition information corresponding to each road type can be accurately obtained. Combined with the speed information, the slip rate of each vehicle tire reflects the interaction between the tire and the road. This information is then combined with the vehicle's curb weight, a key factor influencing vehicle dynamics. These data and information then work together to more comprehensively and accurately simulate the vehicle's driving state under different road conditions, thereby determining the tire model used in the vehicle simulation test platform. This tire model more realistically reflects the characteristics and behavior of tires in actual driving, helping to improve the accuracy and reliability of vehicle simulation testing, providing more effective support for vehicle R&D and performance evaluation, reducing actual testing costs and time, and accelerating the vehicle development process. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0069] Figure 1 A schematic diagram of the architecture corresponding to the method for constructing a vehicle simulation test platform provided in an embodiment of the present application;

[0070] Figure 2 Schematic diagram of the process of constructing the vehicle simulation test platform provided in the embodiment of the present application Figure 1 ;

[0071] Figure 3 Schematic diagram of the process of constructing the vehicle simulation test platform provided in the embodiment of the present application Figure 2 ;

[0072] Figure 4 This is a schematic diagram of the algorithm processing architecture of the convolutional neural network model provided in the embodiment of the present application;

[0073] Figure 5 Schematic diagram of the process of constructing the vehicle simulation test platform provided in the embodiment of the present application Figure 3 ;

[0074] Figure 6 Schematic diagram of the process of constructing the vehicle simulation test platform provided in the embodiment of the present application Figure 4 ;

[0075] Figure 7 A schematic diagram of the maximum longitudinal driving force of the improved rear tire model provided in an embodiment of the present application;

[0076] Figure 8 Schematic diagram of the process of constructing the vehicle simulation test platform provided in the embodiment of the present application Figure 5 ;

[0077] Figure 9 A schematic diagram of the structure of a device for constructing a vehicle simulation test platform provided in an embodiment of the present application;

[0078] Figure 10 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application.

[0079] The above drawings illustrate specific embodiments of the present application, which will be described in more detail below. These drawings and the textual description are not intended to limit the scope of the present application in any way, but rather to illustrate the concepts of the present application to those skilled in the art by reference to specific embodiments. DETAILED DESCRIPTION

[0080] Exemplary embodiments will be described in detail herein, with examples illustrated in the accompanying drawings. In the following description, when referring to the drawings, identical numerals in different figures represent identical or similar elements, unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all embodiments consistent with the present application. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present application, as detailed in the appended claims.

[0081] With the gradual deepening of vehicle intelligent control systems, the demand for vehicle controllers to control vehicles and their tires is becoming increasingly strong. In order to realize the testing and verification of high-level vehicle controller control strategies, a simulation test platform is built to develop vehicle software functions by using a model-in-the-loop or hardware-in-the-loop method for the control strategy. This not only shortens the software development process cycle, but also reduces safety hazards in actual vehicle test scenarios such as extreme acceleration and drive anti-skid.

[0082] The road adhesion coefficient of a vehicle's tires is used to describe the adhesion between the tires and the road, and is also a key factor in determining whether the vehicle can fully exert its dynamic performance.

[0083] Currently, the wheel-to-ground adhesion coefficient parameter needs to be considered in most high-level vehicle safety function control strategies, such as the Anti-lock Braking System (ABS) and Acceleration Slip Regulation (ASR).

[0084] In the existing technology, the road adhesion coefficient estimation methods are mainly divided into three categories: experimental method, model method and fusion method. The experimental method uses sensors such as body sensors, sound waves, cameras, etc. to identify the road or vehicle status, and estimates the road adhesion coefficient on this basis. The model method is based on vehicle dynamics theory and establishes tire force models, such as the Dugoff model and the magic tire formula model, to estimate the road adhesion coefficient based on the μ-s curve slope method. The fusion method combines the experimental and model methods, and improves the accuracy of the estimation of the road adhesion coefficient with the support of big data and intelligent optimization algorithms. This method has a high computational complexity and poses a great challenge to the controller hardware. The complex calculation process needs to be simplified to meet the real-time control requirements.

[0085] However, based on the analysis of the characteristics of the above-mentioned road adhesion coefficient estimation methods, the above-mentioned methods all have the following disadvantages:

[0086] 1. The experimental method has problems such as high sensor accuracy requirements and high difficulty in image recognition algorithms;

[0087] 2, the model method needs to use second-order filtering or Kalman filtering algorithm to filter the collected vehicle body state signals when the vehicle is actually driving, which is not conducive to real-time control of the vehicle;

[0088] 3, the fusion method has not fully played the performance advantages based on image analysis technology and accurate dynamic model, and it is particularly important to reduce the hardware burden of the controller and improve the test ability of the vehicle simulation platform.

[0089] Based on the above-mentioned technical problems, the technical concept of the inventor is as follows: in the process of actually obtaining the related parameters for constructing the vehicle simulation test platform, the road surface condition information corresponding to different road types and some road image data of the corresponding road surface can be constructed in advance, a convolutional neural network model capable of identifying the real-time collected road image data of the vehicle can be trained, and the corresponding target road surface condition information can be obtained, which can represent the road adhesion condition estimation characteristic; further, based on the slip ratio of each tire, the target road surface condition information, and the curb weight of the vehicle, the tire model in the vehicle simulation test platform can be more accurately constructed.

[0090] In this concept, instead of relying on the road adhesion coefficient estimated by a single sensor or image recognition algorithm, the convolutional neural network algorithm is used to obtain the road condition information, and the extended Kalman filter algorithm is used for fusion to estimate the tire slip state, and the test ability of the vehicle simulation platform to the power domain software and its strategy can be determined.

[0091] Based on the above technical concept, Figure 1 The construction method of the vehicle simulation test platform provided by the embodiment of the application corresponds to the architecture schematic diagram as shown in Figure 1 It is shown that it includes: historical road surface data set, real-time road surface photograph, Internet of Things real-time displacement data, longitudinal acceleration sensor, lateral acceleration sensor, and wheel speed sensor.

[0092] The historical road surface data set is used to train the convolutional neural network algorithm, the road surface image data obtained by the real-time road surface photograph is used to input into the trained convolutional neural network algorithm to obtain the current road surface condition, and the peak adhesion coefficient of the road surface is parsed from the current road surface condition;

[0093] The data collected by the Internet of Things real-time displacement data, the longitudinal acceleration sensor, and the lateral acceleration sensor is input into the extended Kalman filter (EKF) state observer to obtain the current actual vehicle speed, and the data obtained by the wheel speed sensor is combined to obtain the slip ratio state of each tire.

[0094] Furthermore, based on the peak road adhesion coefficient and the slip state of each tire, the improved magic formula tire model is determined, specifically including: peak factor calculation, shape factor calculation, tire slip calculation, and longitudinal force calculation function; then, the tire force condition is determined, that is, the maximum longitudinal driving force of the improved tire model (for example, using the improved magic formula tire model), the schematic diagram of which is shown below. Figure 7 .

[0095] For details not disclosed in the above architecture, please refer to the following embodiments.

[0096] The method for constructing a vehicle simulation test platform provided in the embodiment of the present application can be executed by electronic devices such as terminal devices, servers, and computers.

[0097] The following specific embodiments describe in detail the technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0098] Figure 2 Schematic diagram of the process of constructing the vehicle simulation test platform provided in the embodiment of the present application Figure 1 ,like Figure 2 As shown, the method includes:

[0099] Step 21: Acquire road surface image data of a vehicle traveling on at least one road type, speed information of the vehicle on at least one road type, and curb weight of the vehicle;

[0100] In this step, the vehicle-mounted camera is used to take real-time photos of the road surface of different road types to obtain road surface image data corresponding to different road types; and the acceleration sensor (longitudinal and lateral), wheel speed sensor, etc. are used to collect vehicle speed information.

[0101] Optionally, the road surface image data may be image features of the road surface image, that is, each pixel of the road surface image is regarded as a data set of different size values ​​after storage.

[0102] Exemplarily, the at least one road type includes: snowy road surface, wet cobblestone road surface, wet asphalt road surface, and the like.

[0103] Step 22: Input at least one road surface image data into a preset convolutional neural network model to obtain target road surface condition information corresponding to at least one road type;

[0104] In this step, for each road type, the road surface image data corresponding to the road type is input into a preset convolutional neural network model, which can be obtained based on training data and test data, to obtain road surface condition information capable of representing the current road type, denoted as target road surface condition information.

[0105] The training data and the test data include a plurality of historical acquired road surface image data and road surface condition information corresponding to the plurality of historical acquired road surface image data.

[0106] In a possible implementation, the target road surface condition information can include road surface peak adhesion coefficient and other road surface parameters.

[0107] For example, Table 1 is a table of different road surface peak adhesion coefficients provided by the embodiments of the present application, as shown in Table 1:

[0108] Table 1

[0109] Road type Peak adhesion coefficient Icy and snowy roads 0.1-0.2 muddy road 0.2-0.4 Gravel or cobblestone road surface 0.3-0.5 Wet or slippery roads or concrete surfaces 0.4-0.6 Dry asphalt road 0.7-0.8

[0110] Step 23, determining a tire model in the vehicle simulation test platform based on the slip ratio of each tire in the vehicle corresponding to the at least one speed information, the at least one target road surface condition information, and the kerb mass of the vehicle.

[0111] In this step, based on the at least one speed information, the slip ratio of each tire in the vehicle, the at least one target road surface condition information, and the kerb mass of the vehicle are input into a specific algorithm or a preset model construction tool, the influence of the slip ratio, the road surface condition, and the kerb mass on the mechanical properties of the tire is comprehensively considered through mathematical calculation and simulation analysis, the force and deformation behaviors of the tire are modeled, and finally the tire model in the vehicle simulation test platform is determined.

[0112] The method for constructing a vehicle simulation test platform provided in an embodiment of the present application obtains road surface image data of a vehicle traveling on at least one road type, as well as the vehicle's speed information and curb weight for the at least one road type; inputs the at least one road surface image data into a preset convolutional neural network model to obtain target road surface condition information corresponding to the at least one road type; and determines a tire model in the vehicle simulation test platform based on the slip rate of each tire on the vehicle corresponding to the at least one speed information, the at least one target road surface condition information, and the curb weight of the vehicle. In this technical solution, by obtaining road surface image data, speed information, and curb weight for the vehicle traveling on at least one road type, multi-dimensional vehicle driving-related data is collected, and the road surface image data is input into a preset convolutional neural network model. The model's powerful image recognition and feature extraction capabilities are utilized to accurately obtain the target road surface condition information corresponding to each road type. Simultaneously, the slip rate of each tire of the vehicle corresponding to the speed information, which can reflect the interaction between the tire and the road surface, is combined with the vehicle's curb weight, a key factor affecting vehicle dynamics. These data and information then work together to more comprehensively and accurately simulate the vehicle's driving state under different road conditions, thereby determining the tire model in the vehicle simulation test platform. In other words, the tire model can more realistically reflect the characteristics and behavior of the tire during actual driving, help improve the accuracy and reliability of vehicle simulation testing, provide more effective support for vehicle research and development, performance evaluation, etc., reduce actual testing costs and time, and accelerate the vehicle development process.

[0113] Based on the above embodiments, Figure 3 Schematic diagram of the process of constructing the vehicle simulation test platform provided in the embodiment of the present application Figure 2 ,like Figure 3 As shown, the above step 23 may include the following steps:

[0114] Among them, combined Figure 4 , the implementation of this embodiment is described, Figure 4 Schematic diagram of the algorithm processing architecture of the convolutional neural network model provided in the embodiments of the present application.

[0115] Step 31: for each road surface image data, perform convolution processing, maximum pooling, and data flattening processing on the road surface image data in sequence to obtain a fully connected layer corresponding to at least one single-dimensional data group corresponding to the road surface image data;

[0116] The convolution processing is to perform convolution operations in a preset number of image processing channels in the convolutional neural network model. The preset number is obtained based on the RGB values ​​for each single-dimensional data, and the single-dimensional data is used to mark the abstract features of the pixels in the road image data.

[0117] In one possible implementation, for road surface image data of the same road type, 12 image processing channels are generated based on the RGB values ​​(red, green, and blue color intensity). In each channel, the image is subjected to convolution operations, maximum pooling, and data flattening, and then a fully connected layer corresponding to at least one single-dimensional data group is obtained.

[0118] Optionally, (wherein the fully connected layer corresponding to at least one single-dimensional data set includes: a first fully connected layer, a second fully connected layer, and a third fully connected layer), a possible implementation of step 31 may be (combined with Figure 4 ):

[0119] Step 1: Perform convolution processing on the road surface image data to obtain a first feature map set corresponding to the road surface image data;

[0120] In this implementation, the road surface image data is input into the convolutional layer CL1 for preprocessing to obtain a first feature map set corresponding to the road surface image data.

[0121] Among them, a convolution operation is performed on the image (road surface image data) in each image processing channel, and each road image with a pixel size of 64×64 is marked as (64, 64, i), where i = 1, 2…12. The convolution kernel marked as (3, 3, i) is used to perform a convolution operation on the image pixel data on each channel respectively, and the convolution step size is 3. The feature map in each image is extracted, and after adding data calculation bias, the 12 feature maps are superimposed to form the first feature map set FP1. At this time, the pixels of the feature map set are 62×62.

[0122] Step 2: Perform maximum pooling and data flattening on the first feature map set to form the first fully connected layer corresponding to the first single-dimensional data group;

[0123] In this implementation, the ReLU function is used to process the data of the first feature map set FP1, eliminating pixel data below 0. At the same time, to reduce data complexity, the unit matrix with component dimension (2, 2) is used to perform maximum pooling on the first feature map set. At this time, the pixel data size is 31×31, completing the operation of the pooling layer PL1.

[0124] Furthermore, after the pooling layer processing is completed, the integrated image data group is subjected to data flattening processing, that is, the 31×31 image data group is converted into a single-dimensional data group to form a fully connected layer FL1.

[0125] Step 3: Perform maximum pooling and data flattening on the second feature map set in sequence to form a second fully connected layer corresponding to the second single-dimensional data group. The second feature map set is obtained by performing maximum pooling and then convolution on the first feature map set.

[0126] In this implementation, the integrated image data group obtained after the maximum pooling process is performed on the first feature map set is convolved to extract the eigenvalues ​​to obtain the second feature map set FP2.

[0127] Afterwards, the maximum pooling PL2 and data flattening processing are performed using the method in step 2 to form the second fully connected layer FL2 corresponding to the second single-dimensional data group.

[0128] Step 4: Perform maximum pooling and data flattening on the third feature map set in sequence to form a third fully connected layer corresponding to the third single-dimensional data group. The third feature map set is obtained by performing maximum pooling and convolution on the second feature map set.

[0129] In this implementation, the integrated image data set obtained after the maximum pooling process is performed on the second feature map set is convolutionally processed to extract feature values ​​to obtain the third feature map set FP3.

[0130] Afterwards, the maximum pooling PL3 and data flattening processing are performed using the method in step 2 to form the third fully connected layer FL3 corresponding to the third single-dimensional data group.

[0131] Step 32: Based on at least one fully connected layer, the mapping and operation processing of the neural network node input and the hidden layer in the convolutional neural network model are performed to obtain the target data result;

[0132] In this step, after obtaining the first fully connected layer, the second fully connected layer, and the third fully connected layer, the mapping and operation processing with the neural network node input and the hidden layer will eventually calculate a set of data results, which will be recorded as the target data results.

[0133] Step 33: Match the target data result with at least one preset road surface condition information to obtain target road surface condition information corresponding to the road surface image data.

[0134] In this step, the Softmax function is used to match the target data results with different road condition information (e.g., numerical values), and the estimated value with the highest probability of meeting the current road adhesion condition is output as the target road condition information corresponding to the road image data (i.e., Figure 4 road conditions in the medium).

[0135] The method for constructing a vehicle simulation test platform provided in an embodiment of the present application performs convolution processing, maximum pooling, and data flattening processing on each road surface image data in sequence to obtain a fully connected layer corresponding to at least one single-dimensional data group corresponding to the road surface image data; wherein the convolution processing is to perform a convolution operation in a preset number of image processing channels in a convolutional neural network model, and the preset number is obtained based on RGB values. For each single-dimensional data, the single-dimensional data is used to mark the abstract features of the pixels in the road surface image data; based on at least one fully connected layer, the mapping and operation processing of the neural network node input and the hidden layer in the convolutional neural network model are performed to obtain the target data result; by matching the target data result with at least one preset road surface condition information, the target road surface condition information corresponding to the road surface image data is obtained. In this technical solution, convolution processing determines a preset number of image processing channels based on RGB values ​​for convolution operations, which can effectively extract the abstract features of pixels in road images and use the characteristics of convolutional neural networks to capture local information and structure of images, such as identifying road texture, cracks, potholes and other features; the maximum pooling operation reduces the amount of data, reduces computational complexity, improves processing efficiency, and prevents overfitting to a certain extent while retaining the main features; data flattening processing converts multi-dimensional data into a single-dimensional data set to prepare for the subsequent fully connected layer processing. Based on these fully connected layers, through mapping and calculation processing with the neural network node input and hidden layer in the convolutional neural network model, the feature information extracted previously is further integrated and analyzed, and more representative and valuable data is mined to ultimately obtain the target data results. Finally, the target data results are matched with the preset road condition information, and the target road condition information corresponding to the road image data can be accurately obtained.

[0136] On the basis of the above embodiment, for each speed information, the speed information includes: wheel speed information of at least one wheel and acceleration information of the vehicle; then before step 23, the following embodiment may also be performed, namely Figure 5 Schematic diagram of the process of constructing the vehicle simulation test platform provided in the embodiment of the present application Figure 3 ,like Figure 5 As shown, the method may further include the following steps:

[0137] Step 51: Using an extended Kalman filter algorithm, the wheel speed information of at least one wheel and the acceleration information of the vehicle are integrated to obtain the actual speed information of the vehicle;

[0138] Existing technologies offer three methods for calculating a vehicle's actual speed: 1) Using the Galilean coordinate system to establish spatial vectors and coordinates, the current vehicle speed is calculated by differentiating the distance traveled with respect to time. 2) When road adhesion is good, the contact between the driven wheel and the road is considered pure rolling. When driving on real roads, vehicles are typically equipped with wheel speed sensors, which can be used as known quantities for calculating current speed. 3) High-end vehicles are often equipped with acceleration sensors. Using the integral method, the vehicle's acceleration is directly integrated to calculate the vehicle's speed between t1 and t2, as shown in the following equation. This method can also predict future vehicle speeds based on Newton's second law. However, the existing approach has certain limitations. 1) While accurate vehicle speed information can be calculated in a virtual environment, it ignores the relationship between wheel and tire adhesion and the road surface, making it incapable of being directly used in the development of vehicle power domain control strategies. 2) The actual vehicle speed calculation is not accurate enough, making it impossible to represent the actual vehicle speed when the driven wheels are on potholes or icy or snowy roads. 3) Integrating acceleration for a long period of time will result in excessive integration errors, making it unsuitable for determining vehicle speed alone.

[0139] In this step, the extended Kalman filter (EKF) algorithm is used to estimate the current vehicle speed. Its basic principle is to fuse the current wheel speed information with the vehicle body acceleration information based on the dynamic model, and consider the impact of nonlinear characteristics of tire slip on vehicle speed, so that the vehicle dynamics simulation platform is closer to the response state in the actual vehicle driving engineering.

[0140] Optionally, a possible implementation of step 51 may be:

[0141] Step 1: Determine the state equation and observation equation of the observer in the extended Kalman filter algorithm based on the wheel speed information of at least one wheel and the acceleration information of the vehicle, wherein the state equation includes: the vehicle's yaw angular acceleration equation, the center of mass sideslip angular velocity equation, and the vehicle's longitudinal acceleration equation, and the observation equation includes: the center of mass lateral acceleration equation;

[0142] In this implementation, the state equation of the extended Kalman filter observer based on the vehicle dynamics model is:

[0143] The vehicle's yaw angular acceleration equation is:

[0144]

[0145] The equation of the sideslip angular velocity of the center of mass is:

[0146]

[0147] The longitudinal acceleration equation of the vehicle is:

[0148]

[0149] where ω is the body yaw rate; is the lateral acceleration; β is the center of mass side slip angle, is the center of mass side slip rate; a, b are the longitudinal distances from the front and rear axles to the center of mass; k1, k2: are the side slip stiffness of the front and rear tires, respectively; Ix, Iz are the rotational inertia of the vehicle around the x and z axes, respectively; δ is the front wheel steering angle; m is the mass of the vehicle; v x is the longitudinal velocity of the vehicle (determined from wheel speed information); a x is the longitudinal acceleration of the center of mass (i.e., acceleration information).

[0150] The observation equation includes: the center of mass lateral acceleration equation:

[0151]

[0152] where a y is the center of mass lateral acceleration.

[0153] Step 2, according to the general form equation of the observer in the extended Kalman filter algorithm, the state equation, and the observation equation, determine the actual speed information of the vehicle.

[0154] In this implementation, the general form equation of the observer in the extended Kalman filter algorithm is:

[0155]

[0156] y(t) = h(x(t), v(t))

[0157] where x(t) is the state variable; u(t) is the control variable; w(t) is the system excitation noise; v(t) is the measurement noise; y(t) is the measurement output.

[0158] Further, after linearizing the model, the state equation and the observation equation are obtained:

[0159] F(t) formula:

[0160]

[0161] H(t) formula:

[0162]

[0163] Further, the EKF filtering calculation process can be:

[0164] Step 1, prior estimation:

[0165] x(k+1|k) = f(x(k), u(k))

[0166] Step 2, error covariance matrix update:

[0167] P(k+1|k)=φ(k+1|k)P(k|k)φ T (k+1|k)+Q(k+1)

[0168] Step 3. Kalman gain calculation:

[0169] K(k+1)=P(k+1|k)H T (k+1)[H(k+1)P(k+1|k)H T (k+1)+R(k+1)] -1

[0170] Step 4. Calculate the state variables at time k+1:

[0171] x(k+1|k+1)=x(k+1|k)+K(k+1)[Y(k+1)-h(x(k+1,k)u(k+1))]

[0172] Step 5. Correct the error covariance matrix:

[0173] P(k+1|k)=[IK(k+1)H(k+1)]P(k+1|k)

[0174] Where P(k+1|k) is the error covariance prediction matrix at time k+1; K(k+1) is the Kalman gain at time k+1;

[0175] Update the status in real time and go to step 1;

[0176] Step 6. Output the predicted value.

[0177] That is, the actual speed information of the vehicle is obtained.

[0178] Step 52: For each wheel, determine the slip rate of the tire corresponding to the wheel based on the actual speed information and the wheel speed information.

[0179] In this step, the current slip rate of each tire is determined based on the actual speed information of the vehicle. A possible implementation formula is:

[0180]

[0181] Among them, λ cx is the slip rate; v o is the actual speed information of the vehicle; v i It is the wheel speed information of the driving wheel.

[0182] It should be understood that if the tire is a driving wheel, the slip rate of the tire is determined according to the above formula; if the tire is a non-driving wheel, the slip rate of the tire is 0.

[0183] The method for constructing a vehicle simulation test platform provided in an embodiment of the present application employs an extended Kalman filter algorithm to fuse the wheel speed information of at least one wheel with the vehicle's acceleration information to obtain the vehicle's actual speed information. The slip ratio of the tire corresponding to each wheel is then determined based on the actual speed information and the wheel speed information. This technical solution fuses the wheel speed information with the vehicle acceleration information, leveraging the complementarity between these different information sources to reduce potential errors and interference associated with a single information source. This allows for a more accurate estimation of the vehicle's actual speed information, and consequently, an accurate slip ratio, thereby improving the accuracy of subsequent model construction.

[0184] On the basis of the above embodiment, the target road condition information includes: road peak adhesion coefficient; the implementation of step 23 may include the content of the following embodiment, namely: Figure 6 Schematic diagram of the process of constructing the vehicle simulation test platform provided in the embodiment of the present application Figure 4 ,like Figure 6 As shown, the method may include:

[0185] Step 61: Determine a calculation formula for the actual maximum grip of each tire corresponding to the at least one road type based on the peak road adhesion coefficient corresponding to the at least one road type, the slip rate of each tire corresponding to the at least one road type, and the curb weight of the vehicle;

[0186] In this step, for each road type, based on the peak road adhesion coefficient corresponding to the road type, the slip rate of each tire corresponding to the road type, and the curb weight of the vehicle, a calculation formula for the actual maximum grip of each tire corresponding to the road type can be calculated.

[0187] Optionally, a possible implementation of step 61 may be:

[0188] Step 1: For each tire, determine the theoretical maximum longitudinal grip of the tire based on the maximum peak adhesion coefficient among the peak adhesion coefficients of the road surface and the vertical pressure of the tire on the ground;

[0189] In this implementation, the theoretical maximum grip of the tire in longitudinal direction is calculated as:

[0190] D=μ f max ·F Z

[0191] L=μ f min ·F Z

[0192] Where D is the theoretical maximum grip of the tire in longitudinal driving; L is the theoretical minimum grip of the tire in longitudinal driving; μ f max is the maximum peak adhesion coefficient; μ f min is the minimum peak adhesion coefficient; Fz is the vertical pressure of the tire on the ground.

[0193] It should be understood that in this formula, D is a parameter determined based on the maximum peak adhesion coefficient of the road surface and the vertical pressure of the tire on the ground. It can be regarded as a basic coefficient of the longitudinal force of the tire under ideal conditions (corresponding to the maximum peak adhesion coefficient of the road surface). It reflects the basic influence of the road surface and tire vertical load on the longitudinal force, but is not the actual tire longitudinal force.

[0194] Step 2: Determine a calculation formula for the actual maximum grip of the tire corresponding to the at least one road type based on the theoretical maximum grip, the slip rate of the tire corresponding to the at least one road type, and the curb weight of the vehicle.

[0195] In this implementation, for each road type, the maximum grip of the tire corresponding to the road type (F x is the actual maximum grip for longitudinal driving) and the calculation formula is:

[0196]

[0197] Among them, K k ,k, The same is the shape factor, which is the result of data fitting. μx is the proportional coefficient of friction effect.

[0198] It should be understood that: F x The longitudinal grip of the tire under actual conditions is obtained through a series of calculations using the improved magic formula tire model, which takes into account the road adhesion coefficient range (including μ fmax 、μ fmin ), tire slip rate λ cx , and other intermediate parameters (B, C, The maximum longitudinal grip that a tire can actually generate under the combined effect of factors such as the road surface, the ground, and the environment (such as the road surface, etc.) is the mechanical performance value under actual working conditions.

[0199] Step 62: Determine a tire model in the vehicle simulation test platform according to a calculation formula for the actual maximum grip of each tire corresponding to at least one road type.

[0200] In this step, the calculation formula of the actual maximum grip of each tire corresponding to each road type is used to determine the tire model in the vehicle simulation test platform.

[0201] When the weight of a four-wheeled sedan is set to 2200 kg, the present invention is used for calculation, and the simulation effect of the maximum longitudinal driving force of a single tire model under different road conditions is good.

[0202] That is, the tire model is actually used to express the maximum effective grip that the tire can obtain under different conditions, and ultimately the maximum longitudinal driving force of the improved tire model can be obtained. A possible implementation of the maximum longitudinal driving force is as follows: Figure 7 ( Figure 7 As shown in the schematic diagram of the longitudinal maximum driving force of the improved rear tire model provided in the embodiment of the present application, it includes: snow road surface, wet cobblestone road surface, and wet asphalt road surface. The schematic diagram records the corresponding relationship between the slip rate and the longitudinal maximum driving force on the above three road surfaces.

[0203] The method for constructing a vehicle simulation test platform provided in an embodiment of the present application determines a calculation formula for the actual maximum grip of each tire corresponding to at least one road type based on the peak road surface adhesion coefficient corresponding to at least one road type, the slip rate of each tire corresponding to at least one road type, and the curb weight of the vehicle. The tire model in the vehicle simulation test platform is then determined based on the calculation formula for the actual maximum grip of each tire corresponding to at least one road type. This technical solution, which integrates the peak road surface adhesion coefficient, tire slip rate, and vehicle curb weight to determine the calculation formula for the actual maximum grip of the tire, and then determines the tire model for the vehicle simulation test platform based on this formula, can realistically simulate the mechanical properties of the tire, assisting in vehicle research and development, performance optimization, and safety assessment, thereby improving the overall performance and safety of the vehicle.

[0204] Based on the above embodiments, Figure 8 Schematic diagram of the process of constructing the vehicle simulation test platform provided in the embodiment of the present application Figure 5 ,like Figure 8 As shown, the method may include:

[0205] Step 81: Use a vehicle-mounted camera to take real-time photos of the road, establish an image dataset, identify target road condition information based on a convolutional neural network model algorithm, and estimate road condition parameters such as the peak road adhesion coefficient;

[0206] Step 82: Use the vehicle acceleration sensor, wheel speed sensor, etc. to collect the current state of the vehicle, and use the EKF algorithm to fuse the vehicle speed and acceleration signals to obtain the current actual vehicle speed;

[0207] Step 83: Based on the established vehicle dynamics simulation model, the slip rate of each tire is calculated according to the relationship between the current actual vehicle speed and the wheel speed of each tire;

[0208] Step 84: Optimize the tire model in the vehicle simulation test platform based on the slip rate of each tire, road condition parameters, and the curb weight of the vehicle.

[0209] The technical principles and technical effects of the method for constructing a vehicle simulation test platform provided in the embodiment of the present application are similar to those of the above-mentioned embodiment and will not be repeated here.

[0210] Figure 9 A schematic diagram of the structure of a vehicle simulation test platform construction device provided in an embodiment of the present application, such as Figure 9 As shown, the construction device includes:

[0211] an acquisition module 91 for acquiring road surface image data of a vehicle traveling on at least one road type, speed information of the vehicle on at least one road type, and curb weight of the vehicle;

[0212] a processing module 92 for inputting at least one road surface image data into a preset convolutional neural network model to obtain target road surface condition information corresponding to at least one road type;

[0213] The determination module 93 is configured to determine a tire model in the vehicle simulation test platform based on the slip rate of each tire in the vehicle corresponding to at least one speed information, at least one target road condition information, and the curb weight of the vehicle.

[0214] In one or more embodiments, the processing module 92 is specifically configured to:

[0215] For each road surface image data, convolution processing, maximum pooling, and data flattening processing are sequentially performed on the road surface image data to obtain a fully connected layer corresponding to at least one single-dimensional data group corresponding to the road surface image data. The convolution processing is to perform convolution operations in a preset number of image processing channels in the convolutional neural network model. The preset number is obtained based on RGB values. For each single-dimensional data, the single-dimensional data is used to mark the abstract features of the pixels in the road surface image data;

[0216] Based on at least one fully connected layer, the mapping and operation processing between the neural network node input and the hidden layer in the convolutional neural network model is performed to obtain the target data result;

[0217] By matching the target data result with at least one preset road surface condition information, target road surface condition information corresponding to the road surface image data is obtained.

[0218] In one or more embodiments, the processing module 92 sequentially performs convolution, maximum pooling, and data flattening on the road surface image data to obtain a fully connected layer corresponding to at least one single-dimensional data set corresponding to the road surface image data, specifically for:

[0219] Performing convolution processing on the road surface image data to obtain a first feature map set corresponding to the road surface image data;

[0220] Performing maximum pooling and data flattening on the first feature map set in sequence to form a first fully connected layer corresponding to the first single-dimensional data group;

[0221] Performing maximum pooling and data flattening processing on the second feature map set in sequence to form a second fully connected layer corresponding to the second single-dimensional data group, where the second feature map set is obtained by performing maximum pooling processing and then convolution processing on the first feature map set;

[0222] Performing maximum pooling and data flattening processing on the third feature map set in sequence to form a third fully connected layer corresponding to the third single-dimensional data group, where the third feature map set is obtained by performing maximum pooling processing and then convolution processing on the second feature map set;

[0223] The fully connected layers corresponding to the at least one single-dimensional data group include: a first fully connected layer, a second fully connected layer, and a third fully connected layer.

[0224] In one or more embodiments, for each piece of speed information, the speed information includes: wheel speed information of at least one wheel and acceleration information of the vehicle;

[0225] Accordingly, before determining the tire model in the vehicle simulation test platform based on the slip rate of each tire in the vehicle corresponding to the at least one speed information, the at least one target road condition information, and the curb weight of the vehicle, the processing module 92 is further configured to:

[0226] The extended Kalman filter algorithm is used to fuse the wheel speed information of at least one wheel and the acceleration information of the vehicle to obtain the actual speed information of the vehicle;

[0227] For each wheel, the slip rate of the tire corresponding to the wheel is determined based on the actual speed information and the wheel speed information.

[0228] In one or more embodiments, the processing module 92 uses an extended Kalman filter algorithm to fuse the wheel speed information of at least one wheel and the acceleration information of the vehicle to obtain the actual speed information of the vehicle, specifically for:

[0229] Determining a state equation and an observation equation of an observer in an extended Kalman filter algorithm based on wheel speed information of at least one wheel and acceleration information of the vehicle, wherein the state equation includes a yaw acceleration equation, a center of mass sideslip velocity equation, and a longitudinal acceleration equation of the vehicle, and the observation equation includes a center of mass lateral acceleration equation;

[0230] The actual speed information of the vehicle is determined according to the general form equation, state equation, and observation equation of the observer in the extended Kalman filter algorithm.

[0231] In one or more embodiments, the target road surface condition information includes: a peak road surface adhesion coefficient;

[0232] Accordingly, the determination module 93 is specifically configured to:

[0233] Determining a calculation formula for the actual maximum grip of each tire corresponding to the at least one road type based on the peak road adhesion coefficient corresponding to the at least one road type, the slip rate of each tire corresponding to the at least one road type, and the curb weight of the vehicle;

[0234] A tire model in a vehicle simulation test platform is determined according to a calculation formula for the actual maximum grip of each tire corresponding to at least one road type.

[0235] In one or more embodiments, the determination module 93 determines a calculation formula for the actual maximum grip of each tire corresponding to the at least one road type based on the peak road surface adhesion coefficient corresponding to the at least one road type, the slip rate of each tire corresponding to the at least one road type, and the curb weight of the vehicle, specifically for:

[0236] For each tire, the theoretical maximum grip of the tire in longitudinal driving is determined based on the maximum peak adhesion coefficient among the peak adhesion coefficients of the road surface and the vertical pressure of the tire on the ground;

[0237] A calculation formula for the actual maximum grip of the tire corresponding to the at least one road type is determined according to the maximum grip, the slip rate of the tire corresponding to the at least one road type, and the curb weight of the vehicle.

[0238] The device for constructing a vehicle simulation test platform provided in this embodiment can execute the method provided in the above method embodiment. Its implementation principle and technical effects are similar, and are not described in detail in this embodiment.

[0239] Figure 10 This is a schematic diagram of the structure of the electronic device provided in the embodiment of the present application. Figure 10 As shown, the electronic device provided by this embodiment includes: at least one processor 101 and a memory 102. Optionally, the device further includes a communication component 103. The processor 101, the memory 102 and the communication component 103 are connected via a bus 104.

[0240] During the specific implementation process, at least one processor 101 executes the computer-executable instructions stored in the memory 102, so that the at least one processor 101 performs the above method.

[0241] The specific implementation process of the processor 101 can refer to the method embodiments described above, which have similar implementation principles and technical effects, and thus will not be described here again.

[0242] In the above embodiments, it should be understood that the processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), etc. The general-purpose processor can be a microprocessor or any conventional processor, etc. The steps of the method disclosed in the application can be directly embodied as hardware processor execution, or executed by a combination of hardware and software modules in the processor.

[0243] The memory can include a random access memory (RAM), and can also include a non-volatile memory (NVM), such as at least one disk memory.

[0244] The bus can be an industry standard architecture (ISA) bus, a peripheral component (PCI) bus, an extended industry standard architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For the convenience of representation, the bus in the drawings of the present application does not limit to only one bus or one type of bus.

[0245] The present application also provides a computer program product, comprising a computer program, which is executed by a processor to implement the above method.

[0246] The present application also provides a computer readable storage medium, which stores computer execution instructions, and when a processor executes the computer execution instructions, the above method is implemented.

[0247] The above-mentioned readable storage medium can be implemented by any type of volatile or non-volatile memory device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk. The readable storage medium can be any available medium that can be accessed by a general-purpose or special-purpose computer.

[0248] An exemplary readable storage medium is coupled to a processor so that the processor can read information from the readable storage medium and write information to the readable storage medium. Of course, the readable storage medium can also be an integral part of the processor. The processor and the readable storage medium can be located in an application specific integrated circuit (ASIC). Of course, the processor and the readable storage medium can also exist in a device as discrete components.

[0249] The division of units is merely a logical functional division; actual implementations may employ alternative divisions, such as combining or integrating multiple units or components into another system, or omitting or disabling certain features. Furthermore, any direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection between devices or units, either through an interface, electrical, mechanical, or other means.

[0250] Units described as separate components may or may not be physically separate, and components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0251] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0252] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, and other media that can store program code.

[0253] Those skilled in the art will appreciate that all or part of the steps in the above-described method embodiments can be implemented using hardware associated with program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0254] Finally, it should be noted that those skilled in the art will readily identify other embodiments of the present invention after considering the specification and practicing the invention disclosed herein. The present invention is intended to cover any variations, uses, or adaptations of the present invention that follow the general principles of the present invention and include common knowledge or customary techniques in the art not disclosed herein. The present invention is not limited to the precise structure described above and illustrated in the accompanying drawings, and various modifications and variations may be made without departing from the scope thereof. The scope of the present invention is limited solely by the appended claims.

Claims

1. A method for constructing a vehicle simulation test platform, characterized in that: include: Acquiring road surface image data of a vehicle traveling on at least one road type, speed information of the vehicle on the at least one road type, and curb weight of the vehicle; Inputting the at least one road surface image data into a preset convolutional neural network model to obtain target road surface condition information corresponding to the at least one road type; A tire model in the vehicle simulation test platform is determined based on the slip rate of each tire in the vehicle corresponding to the at least one speed information, the at least one target road condition information, and the curb weight of the vehicle.

2. The method according to claim 1, characterized in that Inputting the at least one road surface image data into a preset convolutional neural network model to obtain target road surface condition information corresponding to the at least one road type includes: For each road surface image data, sequentially performing convolution processing, maximum pooling, and data flattening processing on the road surface image data to obtain a fully connected layer corresponding to at least one single-dimensional data group corresponding to the road surface image data, wherein the convolution processing is performed in a preset number of image processing channels in the convolutional neural network model, the preset number being obtained based on RGB values, and for each single-dimensional data, the single-dimensional data is used to mark the abstract features of the pixels in the road surface image data; Obtaining target data results based on mapping and computational processing between the at least one fully connected layer and the neural network node input and hidden layer in the convolutional neural network model; The target road condition information corresponding to the road image data is obtained by matching the target data result with at least one preset road condition information.

3. The method according to claim 2, characterized in that The step of sequentially performing convolution processing, maximum pooling, and data flattening processing on the road surface image data to obtain a fully connected layer corresponding to at least one single-dimensional data group corresponding to the road surface image data includes: Performing the convolution process on the road surface image data to obtain a first feature map set corresponding to the road surface image data; performing the maximum pooling and the data flattening processing on the first feature map set in sequence to form a first fully connected layer corresponding to the first single-dimensional data group; performing the maximum pooling and the data flattening processing on the second feature map set in sequence to form a second fully connected layer corresponding to the second single-dimensional data group, wherein the second feature map set is obtained by performing the maximum pooling processing and then the convolution processing on the first feature map set; performing the maximum pooling and the data flattening processing on a third feature map set in sequence to form a third fully connected layer corresponding to a third one-dimensional data group, wherein the third feature map set is obtained by performing the maximum pooling processing and then the convolution processing on the second feature map set; The fully connected layers corresponding to the at least one single-dimensional data group include: the first fully connected layer, the second fully connected layer, and the third fully connected layer.

4. The method according to any one of claims 1 to 3, characterized in that For each piece of speed information, the speed information includes: wheel speed information of at least one wheel and acceleration information of the vehicle; Accordingly, before determining the tire model in the vehicle simulation test platform based on the slip rate of each tire in the vehicle corresponding to the at least one speed information, the at least one target road condition information, and the curb weight of the vehicle, the method further includes: Using an extended Kalman filter algorithm, the wheel speed information of the at least one wheel and the acceleration information of the vehicle are fused to obtain actual speed information of the vehicle; For each wheel, a slip rate of a tire corresponding to the wheel is determined based on the actual speed information and the wheel speed information of the wheel.

5. The method according to claim 4, characterized in that The method of using an extended Kalman filter algorithm to fuse the wheel speed information of the at least one wheel and the acceleration information of the vehicle to obtain the actual speed information of the vehicle includes: Determining a state equation and an observation equation of an observer in the extended Kalman filter algorithm based on the wheel speed information of the at least one wheel and the acceleration information of the vehicle, the state equation comprising: a yaw angular acceleration equation, a center of mass sideslip angular velocity equation, and a longitudinal acceleration equation of the vehicle, and the observation equation comprising: a center of mass lateral acceleration equation; The actual speed information of the vehicle is determined according to the general form equation of the observer in the extended Kalman filter algorithm, the state equation, and the observation equation.

6. The method according to any one of claims 1 to 3, characterized in that The target road condition information includes: road peak adhesion coefficient; Accordingly, determining a tire model in the vehicle simulation test platform based on the slip rate of each tire in the vehicle corresponding to the at least one speed information, the at least one target road condition information, and the curb weight of the vehicle includes: Determining a calculation formula for the actual maximum grip of each tire corresponding to the at least one road type based on the peak road surface adhesion coefficient corresponding to the at least one road type, the slip rate of each tire corresponding to the at least one road type, and the curb weight of the vehicle; The tire model in the vehicle simulation test platform is determined according to a calculation formula for the actual maximum grip of each tire corresponding to the at least one road type.

7. The method according to claim 6, characterized in that The calculation formula for determining the actual maximum grip of each tire corresponding to the at least one road type based on the peak road surface adhesion coefficient corresponding to the at least one road type, the slip rate of each tire corresponding to the at least one road type, and the curb weight of the vehicle includes: For each tire, determining a theoretical maximum grip of the tire in longitudinal travel based on a maximum peak adhesion coefficient among the road surface peak adhesion coefficients and a vertical pressure exerted by the tire on the ground; A calculation formula for the actual maximum grip of the tire corresponding to the at least one road type is determined based on the theoretical maximum grip, the slip rate of the tire corresponding to the at least one road type, and the curb weight of the vehicle.

8. A device for constructing a vehicle simulation test platform, characterized in that: include: an acquisition module, configured to acquire road surface image data of a vehicle traveling on at least one road type, speed information of the vehicle on the at least one road type, and curb weight of the vehicle; a processing module, configured to input the at least one road surface image data into a preset convolutional neural network model to obtain target road surface condition information corresponding to the at least one road type; A determination module is used to determine a tire model in the vehicle simulation test platform based on the slip rate of each tire in the vehicle corresponding to the at least one speed information, the at least one target road condition information, and the curb weight of the vehicle.

9. An electronic device, characterized in that: include: Memory, processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory, so that the processor performs the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer-executable instructions, which are used to implement the method according to any one of claims 1 to 7 when executed by a processor.

Images