A method for estimating vertical load of intelligent tire

Abstract

The present invention provides an intelligent tire vertical load estimation method, which first collects the tire's operating data under different actual vehicle working conditions and stores them as samples; then learns the samples in combination with a training function to obtain a vertical load estimation model; then imports the tire's ground contact information into the vertical load estimation model to obtain an estimated value of the vertical load. The present invention obtains a vertical load estimation model by collecting the vehicle's operating data under different working conditions and combining it with a training function, imports the model into a real-time controller, and realizes real-time estimation of the vertical load by real-time collection of tire ground contact, wheel speed, and tire pressure information.

Description

Technical Field

[0001] The present invention relates to the field of automobile measurement and control technology, and in particular to a method for estimating vertical load of an intelligent tire. Background Art

[0002] With the guidance of the state on the development direction of the "New Four Modernizations" of the automobile industry, intelligence and networking have also created new development demands for the tire industry. Tire technology integrates advanced sensing technology, signal conditioning technology, real-time communication technology, etc., which can sense and collect tire stress, tire wear and road conditions, and help improve vehicle dynamics control systems and improve vehicle control and driving safety. In order to improve driving safety, handling stability and fuel economy during driving, it is necessary to obtain the vertical load of the tire with high precision and feed it back to the electronic control unit, thereby ensuring driving safety to a greater extent. Summary of the invention

[0003] The object of the present invention is to provide a tire development test system and to achieve estimation of vehicle vertical load based on the system.

[0004] To achieve the above object, the present invention adopts the following technical solution:

[0005] A method for estimating vertical load of an intelligent tire comprises the following steps:

[0006] Collect the operating data of tires under different actual vehicle working conditions and store them as samples;

[0007] The samples are studied in combination with the training function to obtain a vertical load estimation model;

[0008] The tire contact information is imported into the vertical load estimation model to obtain an estimated value of the vertical load.

[0009] Furthermore, the vertical load estimation model is implemented by the following steps:

[0010] a. Define the input and output variables of the model and select the training set and test set;

[0011] b. Preprocess the operation data;

[0012] c. Select the best model parameters through grid search and cross-validation algorithms;

[0013] d. Select the ε-Support Vector Regression (ε-SVR) regression algorithm, and select the Gaussian radial basis (RBF) kernel as the kernel function of the algorithm. Use the optimal parameters to learn the samples to obtain the SVM model. The expression of the kernel function is as follows:

[0014]

[0015] Where: X represents the sample input vector, X c represents the center vector of the kernel function, and σ is the width parameter of the kernel function.

[0016] e. Use the SVM model for regression prediction.

[0017] Preferably, the preprocessing of the running data adopts a normalized algorithm, and the formula is as follows:

[0018]

[0019] Among them: X represents the original data before normalization; Y represents the data after normalization; X min Indicates the minimum value of the sample data; X max Indicates the maximum value of the sample data.

[0020] Furthermore, the smart tire includes a tire and a plurality of PVDF piezoelectric film sensors, a plurality of three-axis acceleration sensors and a temperature and pressure integrated sensor mounted on the tire.

[0021] Furthermore, the contact information of the tire includes a contact longitudinal axis length, a contact transverse axis length, a contact pressure, and a contact area.

[0022] It can be seen from the above technical scheme that the present invention obtains a vertical load estimation model by collecting the vehicle's operating data under different working conditions and combining it with a training function, and imports the model into a real-time controller. By real-time collection of tire contact, wheel speed, and tire pressure information, real-time estimation of the vertical load is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 A flowchart of the specific steps for building a vertical load estimation model;

[0024] Figure 2 Schematic diagram of the vertical load estimation model;

[0025] Figure 3-5 This is a schematic diagram of the estimated effect of the vertical load estimation model; DETAILED DESCRIPTION

[0026] A preferred embodiment of the present invention is described in detail below with reference to the accompanying drawings.

[0027] like Figure 1 The intelligent tire vertical load estimation method shown includes the following steps:

[0028] S1. Collecting the operating data of the tire under different actual vehicle working conditions and storing it as samples;

[0029] S2, learning the samples in combination with the training function to obtain a vertical load estimation model;

[0030] S3. Importing the tire contact information into the vertical load estimation model to obtain an estimated value of the vertical load.

[0031] Specifically, the vertical load estimation model described in this preferred embodiment is implemented by the following steps:

[0032] a. Define the input and output variables of the model and select the training set and test set;

[0033] b. Preprocess the operation data;

[0034] c. Select the best model parameters through grid search and cross-validation algorithms;

[0035] d. Select the ε-Support Vector Regression (ε-SVR) regression algorithm, and select the Gaussian radial basis (RBF) kernel as the kernel function of the algorithm. Use the optimal parameters to learn the sample data to obtain the SVM model. The expression of the kernel function is as follows:

[0036]

[0037] Where: X represents the sample input vector, X c represents the center vector of the kernel function, and σ is the width parameter of the kernel function.

[0038] e. Use the SVM model for regression prediction.

[0039] In order to avoid the influence of different input element size ranges on the output contribution and improve the calculation speed of the model, this preferred embodiment uses a normalized algorithm to preprocess the running data. The formula is as follows:

[0040]

[0041] Among them: X represents the original data before normalization; Y represents the data after normalization; X min Indicates the minimum value of the sample data; X max Indicates the maximum value of the sample data.

[0042] The smart tire described in this preferred embodiment includes a tire and a plurality of PVDF piezoelectric film sensors, a plurality of three-axis acceleration sensors and a temperature and pressure integrated sensor installed on the tire. These sensors can collect the tire's ground contact information in real time. The model is imported into a real-time controller, and the tire's ground contact, wheel speed and tire pressure information are collected in real time to achieve real-time estimation of the vertical load. The tire's ground contact information includes the ground contact longitudinal axis length, ground contact transverse axis length, ground contact pressure and ground contact area.

[0043] like Figure 2 As shown, the vertical load estimation model described in this preferred embodiment is obtained by learning sample data based on the LibSVM toolbox using the svmtrain function. The vertical load estimate obtained by inputting sample data is determined by the svmpredict function, and the prediction accuracy of the model is compared with the actual output value of the sample data. At the same time, the grid search and cross-validation algorithm optimization model is adopted. The idea of ​​this optimization method is to verify all possible parameter pairs, gradually approach the objective function to find the global optimal solution, and thus train the best model. At the same time, the SVM function model has built-in sub-models for multiple working conditions, so the selection of samples needs to cover various test conditions to the maximum extent, so that the model can be fully learned, thereby improving the prediction accuracy of the model.

[0044] like Figure 3-5 As shown in the figure, under three working conditions, namely 80kph-10kph straight braking, 10kph-80kph straight braking and steady left turn with fixed radius (40m), the actual vehicle test was carried out through the vertical load estimation model. In the figure, FL represents the left front wheel and FR represents the right front wheel. It can be seen from the figure that the correlation coefficient between the vertical load estimation value of each wheel position under the three working conditions and the test is greater than 92%, and the mean square error is less than 0.01, which has high accuracy.

[0045] The above-described embodiments are merely descriptions of preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. A method for estimating vertical load of an intelligent tire, characterized in that: The following steps are involved: Collect the operating data of tires under different actual vehicle working conditions and store them as samples; The smart tire comprises a tire and a plurality of PVDF piezoelectric film sensors, a plurality of triaxial acceleration sensors and a temperature and pressure integrated sensor mounted on the tire; The samples are learned in combination with the training function to obtain a vertical load estimation model, which is implemented by the following steps: a. Define the input and output variables of the model and select the training set and test set; b. Preprocess the operation data; c. Select the best model parameters through grid search and cross-validation algorithms; d. Select the ε-Support Vector Regression (ε-SVR) regression algorithm, and select the Gaussian radial basis (RBF) kernel as the kernel function of the algorithm. Use the optimal parameters to learn the samples to obtain the SVM model. The expression of the kernel function is as follows: Where: X represents the sample input vector, X c represents the center vector of the kernel function, σ is the width parameter of the kernel function; e. Use SVM model for regression prediction; Importing tire contact information into the vertical load estimation model to obtain an estimated vertical load; The preprocessing of running data adopts the normalized algorithm, and the formula is as follows: Among them: X represents the original data before normalization; Y represents the data after normalization; X min Indicates the minimum value of the sample data; X max Indicates the maximum value of the sample data; The vertical load estimation model is imported into the real-time controller, and the vertical load is estimated in real time by collecting the tire contact information, wheel speed and tire pressure information in real time; The tire contact information includes contact longitudinal axis length, contact transverse axis length, contact pressure, and contact area.

Images