Torque sensor offset acquisition method and system
By initializing the torque estimation model and optimizing it using the recursive least squares method, combined with a closed-loop adjustment algorithm, the problem of accurately obtaining the torque sensor offset value in the online electric power steering system was solved, improving the torque sensor offset tracking accuracy and response speed, and ensuring the stability and consistency of the steering system.
Patent Information
- Application Number
- CN202410607449.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-05-16
- Publication Date
- 2025-11-18
AI Technical Summary
In online electric power steering systems, the measured torque value of the steering wheel torque sensor deviates from the actual torque value, resulting in inconsistent steering feel. Existing technology struggles to accurately track and compensate for the torque sensor's deviation.
By initializing the torque estimation model, the model parameters are learned using the steering wheel angle and torque sensor measurements. The model parameters are then optimized using the recursive least squares method, and a closed-loop adjustment algorithm is used to improve the tracking speed and accuracy, while the offset of the torque sensor is obtained.
This improves the tracking accuracy and response speed of the torque sensor offset value, ensuring the stability and consistency of the steering system.
Smart Images
Figure CN120970898A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of vehicle steering control, in particular to a torque sensor offset acquisition method and system. BACKGROUND
[0002] In a steer-by-wire electric power steering system, there is no mechanical connection between the steering wheel and the gear, and a load motor is needed to simulate the steering force. In order to keep the consistency of the steering feeling when rotating the steering wheel clockwise and counterclockwise, it is necessary to track and compensate the steering wheel torque sensor offset value in real time. SUMMARY
[0003] In view of the problems in the prior art, the purpose of the present application is to provide a torque sensor offset acquisition method and system to improve the accuracy of torque sensor offset acquisition and improve the tracking accuracy of the steering wheel torque sensor offset value.
[0004] The first aspect of the present application provides a torque sensor offset acquisition method, comprising:
[0005] Initializing the model parameters of a torque estimation model, the input of the torque estimation model being the steering wheel angle and the output being the estimated torque value;
[0006] Obtaining the current steering wheel angle value and the measured torque value currently collected by the torque sensor arranged at the steering wheel;
[0007] Obtaining the estimated torque value according to the steering wheel angle value and the torque estimation model;
[0008] Learning the model parameters of the torque estimation model according to the measured torque value and the estimated torque value;
[0009] Determining the offset of the torque sensor according to the learned model parameters of the torque estimation model.
[0010] In some embodiments, learning the model parameters of the torque estimation model according to the measured torque value and the estimated torque value comprises the following steps:
[0011] Calculating the sum of squared errors according to the measured torque value and the estimated torque value, and learning the model parameters of the torque estimation model based on the recursive least squares method according to the sum of squared errors.
[0012] In some embodiments, after obtaining the estimated torque value according to the steering wheel angle value and the torque estimation model, the following steps are further included:
[0013] Calculating the difference value according to the estimated torque value and the measured torque value;
[0014] determining whether a preset coefficient matrix adjustment condition is met according to the difference value;
[0015] If yes, adjusting a coefficient matrix of the recursive least square method to improve tracking speed of the recursive least square method.
[0016] In some embodiments, determining whether a preset coefficient matrix adjustment condition is met according to the difference value includes the following steps:
[0017] calculating an average value of m continuous difference values, and when the average value is greater than a preset difference threshold value, a counter is incremented, m being an integer greater than 1;
[0018] When a value of the counter is greater than a preset counter threshold value, it is determined that the preset coefficient matrix adjustment condition is met.
[0019] In some embodiments, after the model parameters of the torque estimation model are initialized, the following steps are further included:
[0020] determining whether a torque sensor offset learning condition is met;
[0021] If yes, performing a torque sensor offset learning process.
[0022] In some embodiments, when the following conditions are met, it is determined that the torque sensor offset learning condition is met:
[0023] a current learning enable switch has been turned on;
[0024] a steering wheel angle, a steering wheel torque and a steering wheel speed are respectively within a corresponding preset allowable range; and
[0025] the steering wheel is in a state of not being driven by a user and a duration of the state is greater than a preset time threshold value.
[0026] In some embodiments, determining the offset of the torque sensor according to the learned model parameters of the torque estimation model includes the following steps:
[0027] setting an input of the learned torque estimation model to zero, and obtaining an output value of the torque estimation model as the offset of the torque sensor.
[0028] The second aspect of the present application further provides a torque sensor offset acquisition system for implementing the torque sensor offset acquisition method of the first aspect, the system comprising:
[0029] a parameter initialization module for initializing model parameters of a torque estimation model, an input of the torque estimation model being a steering wheel angle and an output being an estimated torque value;
[0030] a data collection module configured to obtain a current steering wheel angle value and a measured torque value currently collected by a torque sensor arranged at the steering wheel;
[0031] a parameter learning module configured to obtain an estimated torque value according to the steering wheel angle value and the torque estimation model, and learn model parameters of the torque estimation model according to the measured torque value and the estimated torque value;
[0032] an offset obtaining module configured to determine the offset of the torque sensor according to the learned model parameters of the torque estimation model.
[0033] In some embodiments, the parameter learning module is configured to calculate a sum of squared errors according to the measured torque value and the estimated torque value, and learn the model parameters of the torque estimation model based on a recursive least square method according to the sum of squared errors.
[0034] In some embodiments, the system further comprises a precision adjustment module configured to calculate a difference value according to the estimated torque value and the measured torque value, determine whether a preset coefficient matrix adjustment condition is met or not according to the difference value, and if yes, adjust a coefficient matrix of the recursive least square method to improve a tracking speed of the recursive least square method.
[0035] The torque sensor offset obtaining method and system provided by the present application has the following advantages:
[0036] By using the torque sensor offset obtaining method provided by the present application, the model parameters of the torque estimation model are learned based on the steering wheel angle value and the measured torque value of the steering wheel torque sensor, so that the parameters of the torque estimation model are more consistent with the actual scene, the offset of the torque sensor calculated according to the learned torque estimation model is also more accurate, and the tracking precision of the steering wheel torque sensor offset value is higher. BRIEF DESCRIPTION OF DRAWINGS
[0037] Other features, objects, and advantages of the present application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings.
[0038] Figure 1 is a flowchart of a torque sensor offset obtaining method according to an embodiment of the present application;
[0039] Figure 2 is a flowchart of determining whether the torque sensor offset learning condition is met or not according to an embodiment of the present application;
[0040] Figure 3 is a flowchart of adjusting the coefficient matrix of the closed loop adjustment algorithm according to an embodiment of the present application;
[0041] Figure 4is a contrast schematic diagram of tracking effects of offset acquisition methods of two embodiments of the application;
[0042] Figure 5 is a structural block diagram of a torque sensor offset acquisition system of an embodiment of the application. DETAILED DESCRIPTION
[0043] Example implementations will now be described more fully with reference to the accompanying drawings. Example implementations can be implemented in any number of manners, and are not limited to the embodiments described herein; rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the concept of example implementations to those skilled in the art. Like reference numerals in the drawings denote like or similar structures, and thus repeated description thereof will be omitted. In the specification, "or" or "and" can mean "and" or "or". Although the terms "upper", "lower", "between", and the like can be used in this specification to describe different example features and elements of the application, these terms are used herein solely for convenience, e.g., based on the examples shown in the drawings. Nothing in this specification should be construed as requiring a particular three-dimensional orientation of a structure in order to fall within the scope of the application. Although "first" or "second" or the like are used in this specification to denote certain features, this is merely for the purpose of representation of action, and not as a limitation on the number and importance of the specific features.
[0044] In an online control electric power steering system, due to the hardware limitations of the steering wheel torque sensor itself and the external disturbances that can be affected in the use scene, the measured torque value of the steering wheel torque sensor will have a certain offset from the actual torque value of the steering wheel. In order to better track the real torque of the steering wheel, the offset of the steering wheel torque sensor needs to be acquired.
[0045] To solve the technical problems in the prior art, the application provides a torque sensor offset acquisition method for acquiring the offset value of a torque sensor arranged at a steering wheel in an online control electric power steering system, so as to more accurately track the torque sensor offset value of the steering wheel. As shown in the figure, in an embodiment, the torque sensor offset acquisition method comprises the following steps: Figure 1
[0046] S100: initialize the model parameters of the torque estimation model, the input of the torque estimation model being the steering wheel angle and the output being the estimated torque value;
[0047] Here, the model parameters of the torque estimation model can be the model parameters stored last time extracted from the memory. Specifically, after the vehicle is started and the torque sensor is started, the model parameters stored in the memory are acquired, and the model parameters of the torque estimation model are initialized.
[0048] The torque estimation model is a model for calculating a corresponding estimated torque value based on the steering wheel angle, for example, a torque estimation model is Y=A1*sin(X)+A2*cos(X)+B, where Y is the estimated torque value, X is the steering wheel angle, A1, A2 and B are model parameters of the torque estimation model;
[0049] S200: obtaining a current steering wheel angle value and a measured torque value currently collected by a torque sensor arranged at the steering wheel;
[0050] The steering wheel angle value can be obtained by an angle sensor arranged at the steering wheel;
[0051] S300: obtaining an estimated torque value according to the steering wheel angle value and the torque estimation model;
[0052] Specifically, the steering wheel angle value is taken as the input of the torque estimation model, and the estimated torque value output by the torque estimation model is obtained. Taking the above torque estimation model Y=A1*sin(X)+A2*cos(X)+B as an example, the steering wheel angle value is taken as X, and the calculated Y is the estimated torque value;
[0053] S400: learning the model parameters of the torque estimation model according to the measured torque value and the estimated torque value;
[0054] S500: determining the offset of the torque sensor according to the learned model parameters of the torque estimation model.
[0055] After the step S500, if the vehicle is turned off and the torque sensor is powered off, the learned model parameters of the torque estimation model are stored in the memory for initializing the model parameters of the torque estimation model when the vehicle is started next time.
[0056] By using the offset acquisition method of the present application, the model parameters of the torque estimation model are learned based on the steering wheel angle value and the measured torque value of the steering wheel torque sensor, so that the parameters of the torque estimation model are more consistent with the actual scene, the offset of the torque sensor calculated according to the learned torque estimation model is also more accurate, and the tracking accuracy of the steering wheel torque sensor offset value is higher.
[0057] In this embodiment, after the step S100 of initializing the model parameters of the torque estimation model, the following steps are further included:
[0058] determining whether the current meets the torque sensor offset learning condition;
[0059] If yes, the torque sensor offset learning process is performed, i.e., the offset learning process of steps S200-S500 is performed;
[0060] If not, the torque sensor offset learning process will not be executed.
[0061] In this embodiment, the torque sensor offset learning condition is determined to be met when all of the following conditions are satisfied: the current learning enable switch is turned on; the steering wheel angle, steering wheel torque, and steering wheel speed are within their respective preset allowable ranges; and the steering wheel is in a state where it is not driven by the user and the duration of this state is greater than a preset time threshold.
[0062] like Figure 2 The diagram illustrates an implementation process for determining whether the torque sensor offset learning conditions are met. Figure 2 The execution order of each step is merely an example and not intended to limit the scope of protection. Specifically, determining whether the torque sensor offset learning conditions are met includes the following steps:
[0063] S110: Determine whether the current learning enable switch is on;
[0064] If so, continue to S120: determine whether the steering wheel angle, steering wheel torque and steering wheel speed are within their respective preset allowable ranges;
[0065] Specifically, preset allowable ranges for steering wheel angle, steering wheel torque, and steering wheel speed are set respectively. Here, the preset allowable ranges for angle, torque, and speed are all small numerical ranges close to 0. When the steering wheel angle, steering wheel torque, and steering wheel speed are within their respective preset allowable ranges, the steering wheel is basically stationary. The specific numerical ranges can be set as needed.
[0066] If so, continue to S130: determine whether the steering wheel is in a state where it is not being driven by the user; that is, determine whether the user has not placed their hands on the steering wheel.
[0067] If so, continue to S140: determine whether the duration of the steering wheel being in a state without user driving force is greater than a preset time threshold; this step is mainly used for de-shaking processing to ensure that the steering wheel is currently in a stationary state and the user does not operate the steering wheel for a period of time. The length of the preset time threshold can be set as needed. For example, when the preset time threshold is set to a longer value, the number of times offset learning is initiated is less. When the preset time threshold is set to a shorter value, the number of times offset learning is initiated is more.
[0068] If so, determine that the current conditions for torque sensor offset learning are met, and execute the torque sensor offset learning process;
[0069] If the determination in any of the steps S110, S120, S130, S140 is negative, it is determined that the current does not meet the torque sensor offset learning condition, and the torque sensor offset learning process is not performed.
[0070] In this embodiment, the step S300 of learning the model parameters of the torque estimation model according to the measured torque value and the estimated torque value comprises the following steps:
[0071] The sum of squared errors is calculated according to the measured torque value and the estimated torque value, and the model parameters of the torque estimation model are learned based on the recursive least squares method according to the sum of squared errors.
[0072] The recursive least squares method (RLS) is an algorithm applied in online learning and adaptive control, which is used to estimate parameters so that the sum of squared errors between the predicted values of one or more output variables and the actual measured values is minimized. In this embodiment, taking the torque estimation model Y = A1*sin(X) + A2*cos(X) + B as an example, the measured torque value is taken as the actual measured value, the estimated torque value is taken as the predicted value, and the model parameters A1, A2 and B are learned by the recursive least squares method to minimize the sum of squared errors.
[0073] In order to further improve the accuracy and response speed of the torque offset value tracking, after the step S300 of obtaining the estimated torque value according to the steering wheel angle value and the torque estimation model, the step of adjusting the coefficient matrix of the recursive least squares method in a closed loop is further included, so that the model parameters are learned more quickly when the offset value changes greatly, and fast convergence is achieved.
[0074] As shown in the figure, in this embodiment, the coefficient matrix of the recursive least squares method is adjusted in a closed loop by the following steps: Figure 3
[0075] S610: calculating the deviation according to the estimated torque value and the measured torque value;
[0076] S620: judging whether the preset coefficient matrix adjustment condition is met according to the deviation;
[0077] If yes, S630: adjusting the coefficient matrix of the recursive least squares method to improve the tracking speed of the recursive least squares method. Here, the coefficient matrix adjustment condition corresponds to the tracking accuracy condition of the offset value. When the deviation meets the preset coefficient matrix adjustment condition, it means that the current torque sensor offset value tracking accuracy is low, and the coefficient matrix of the recursive least squares method needs to be adjusted.
[0078] AsFigure 3 As shown, the step S620: judging whether the preset coefficient matrix adjustment condition is met according to the difference value, comprises the following steps:
[0079] S621: calculating the average value of m continuous difference values, m is an integer greater than 1;
[0080] The value of m can be set as required, for example, 5, 10, 15, etc.
[0081] Here, the average value of m continuous difference values can be that each m continuous difference value is taken as a group, there is no intersection between adjacent groups, the average value of m continuous difference values of each group is calculated, or each time a difference value is calculated through step S620, the average value of the difference value and the previous m-1 difference values is calculated.
[0082] S622: judging whether the average value is greater than the preset difference threshold value;
[0083] Here, the value of the preset difference threshold value can be set as required. The lower the preset difference threshold value is set, the lower the starting condition of the closed-loop adjustment algorithm coefficient matrix will be. Correspondingly, the higher the preset difference threshold value is set, the higher the requirement for the offset value tracking accuracy will be, that is, the higher the starting condition of the closed-loop adjustment algorithm coefficient matrix will be.
[0084] If no, the coefficient matrix of the recursive least square method is not adjusted.
[0085] If yes, S623: the counter is accumulated.
[0086] S624: judging whether the value of the counter is greater than the preset counter threshold value;
[0087] Here, the value of the preset counter threshold value can be set as required, for example, 10, 15, 20, etc. The size of the preset counter threshold value also affects the starting condition of the closed-loop adjustment algorithm coefficient matrix, that is, the higher the preset counter threshold value is set, the higher the starting condition of the closed-loop adjustment algorithm coefficient matrix will be, and the lower the preset counter threshold value is set, the higher the starting condition of the closed-loop adjustment algorithm coefficient matrix will be.
[0088] If no, the coefficient matrix of the recursive least square method is not adjusted, the current coefficient matrix of the recursive least square method is kept unchanged, and the step of offset value learning is normally executed.
[0089] If yes, it is determined that the preset coefficient matrix adjustment condition is met, and the step S630 of adjusting the coefficient matrix of the recursive least square method to improve the tracking speed of the recursive least square method is continued.
[0090] Specifically, the recursive formula of the recursive least square method is as follows:
[0091]
[0092]
[0093]
[0094] in, and Let y represent the model parameters of the torque estimation model at time k and time k-1, respectively. k This represents the measured torque value of the torque sensor at time k. P represents the steering wheel angle at time k. k and P k-1 Let K represent the coefficient matrices at time k and time k-1, respectively. k This represents the correction gain at time k.
[0095] When k=1, that is, when the recursive least squares method is initially executed, the coefficient matrix P is set empirically. k-1 That is, P0, where each value in the coefficient matrix P0 is an initial value. Based on this coefficient matrix P0, K1 can be calculated, and then... P1 can be calculated based on K1, and used for the calculation of K2, thus continuing the calculation in subsequent time steps. As the recursive least squares method continues to learn, the values in the coefficient matrix P0 will become smaller and smaller, the convergence speed will slow down, and the algorithm's response speed and accuracy will deteriorate. When the preset coefficient matrix adjustment conditions are met, it is necessary to increase the values in the coefficient matrix P0 to improve the convergence speed, thereby improving the algorithm's response speed and accuracy. Increasing the values in the coefficient matrix P0 can be done by increasing each value to its initial value, or by increasing it to its initial value minus a preset coefficient value. The size of the preset coefficient value can be adjusted as needed. The smaller the preset coefficient value is set, the faster the convergence speed after adjustment. When the preset coefficient value is set to 0, it is equivalent to resetting the values in the coefficient matrix to their initial values. The initial value of the coefficient matrix can be set as needed. For example, setting the initial value to a larger value results in a faster convergence speed at the point where the algorithm is learning.
[0096] In this embodiment, step S500, determining the offset of the torque sensor based on the learned model parameters of the torque estimation model, includes the following steps:
[0097] The input of the learned torque estimation model is set to zero, and the output value of the torque estimation model is obtained as the offset of the torque sensor. Taking the above torque estimation model Y=A1*sin(X)+A2*cos(X)+B as an example, when the steering wheel angle X is set to 0, Y=A2+B, and the value of A2+B at this time is the offset value of the torque sensor that has been learned.
[0098] Figure 4 A comparative diagram showing the tracking effect of the offset acquisition method of two embodiments of the present application is shown. The first embodiment is a method of directly obtaining the torque offset value using the above steps S100-S500 without using the closed-loop adjustment process of steps S610-S630, and the second embodiment is a method of using the above steps S100-S500 and adding the closed-loop adjustment process of steps S610-S630. As shown in Figure 4 , the real torque sensor offset value is divided into three sections from left to right, the first section is the real torque sensor offset value is 0.22Nm, the second section is the real torque sensor offset value is 0.83Nm, and the third section is the real torque sensor offset value is 0.31Nm. As can be seen from Figure 4 , the first embodiment of the present application realizes the learning and tracking of the real torque sensor offset in the first section, which can meet the use requirements of some scenes, but the dynamic response of the torque offset value tracking will be affected as the learning time increases. The second embodiment of the present application has higher tracking accuracy of the real torque sensor offset value, and when the real torque sensor offset value changes greatly, the dynamic response of the offset acquisition method of the second embodiment is higher, which can adapt to more use requirements of various scenes. Both of the two embodiments belong to the protection scope of the present application.
[0099] As shown in Figure 5 , the second aspect of the present application further provides a torque sensor offset acquisition system for implementing the torque sensor offset acquisition method of the first aspect, the system comprising:
[0100] A parameter initialization module M100 is configured to initialize the model parameters of the torque estimation model, the input of the torque estimation model being the steering wheel angle and the output being the estimated torque value;
[0101] A data acquisition module M200 is configured to acquire the current steering wheel angle value and the measured torque value currently acquired by the torque sensor arranged at the steering wheel;
[0102] A parameter learning module M300 is configured to acquire the estimated torque value according to the steering wheel angle value and the torque estimation model, and learn the model parameters of the torque estimation model according to the measured torque value and the estimated torque value. In this embodiment, the parameter learning module is configured to calculate the sum of squared errors according to the measured torque value and the estimated torque value, and learn the model parameters of the torque estimation model based on the recursive least squares method according to the sum of squared errors;
[0103] The offset obtaining module M400 is configured to determine the offset of the torque sensor according to the model parameters of the learned torque estimation model. In this embodiment, the offset obtaining module M400 is configured to set the input of the learned torque estimation model to zero, and obtain the output value of the torque estimation model as the offset of the torque sensor.
[0104] In this embodiment, the torque sensor offset obtaining system further comprises a condition judging module configured to judge whether the current meets the torque sensor offset learning condition, and if yes, execute the torque sensor offset learning process, and if no, do not execute the torque sensor offset learning process. The condition judging module judges whether the current meets the torque sensor offset learning condition, for example, by using the steps S110-S140 shown in FIG. 1. Figure 2
[0105] In this embodiment, the torque sensor offset obtaining system further comprises a precision adjusting module configured to calculate a difference value according to the estimated torque value and the measured torque value, judge whether the current meets the preset coefficient matrix adjustment condition according to the difference value, and if yes, adjust the coefficient matrix of the recursive least square method to improve the tracking speed of the recursive least square method. The precision adjusting module judges whether the current meets the preset coefficient matrix adjustment condition according to the difference value, for example, by using the embodiment of step S620 shown in FIG. 6. Figure 3
[0106] The specific functions of each module of the torque sensor offset obtaining system can be realized by using the embodiments of the corresponding steps in the torque sensor offset obtaining method described above, and will not be repeated here.
[0107] The above is a further detailed description of the present application in combination with specific preferred embodiments, and the specific implementation of the present application cannot be limited to these descriptions. For ordinary skilled persons in the technical field to which the present application belongs, several simple deductions or replacements can be made without departing from the concept of the present application, and all of them should be regarded as falling within the protection scope of the present application.
Claims
1. A method for obtaining torque sensor offset, characterized in that, include: Initialize the model parameters of the torque estimation model, where the input of the torque estimation model is the steering wheel angle and the output is the estimated torque value; Obtain the current steering wheel angle value and the measured torque value currently collected by the torque sensor located at the steering wheel; The estimated torque value is obtained based on the steering wheel angle value and the torque estimation model; Based on the measured torque value and the estimated torque value, the model parameters of the torque estimation model are learned; The offset of the torque sensor is determined based on the model parameters of the learned torque estimation model.
2. The torque sensor offset acquisition method according to claim 1, characterized in that, Based on the measured torque value and the estimated torque value, the model parameters of the torque estimation model are learned, including the following steps: The sum of squared errors is calculated based on the measured torque value and the estimated torque value, and the model parameters of the torque estimation model are learned based on the sum of squared errors using the recursive least squares method.
3. The torque sensor offset acquisition method according to claim 2, characterized in that, After obtaining the estimated torque value based on the steering wheel angle value and the torque estimation model, the following steps are also included: Calculate the difference between the estimated torque value and the measured torque value; Based on the difference, determine whether the preset coefficient matrix adjustment conditions are met; If so, adjust the coefficient matrix of the recursive least squares method to improve the tracking speed of the recursive least squares method.
4. The torque sensor offset acquisition method according to claim 3, characterized in that, Determining whether the preset coefficient matrix adjustment conditions are met based on the difference includes the following steps: Calculate the average of m consecutive differences, and increment the counter when the average is greater than a preset difference threshold, where m is an integer greater than 1; When the value of the counter is greater than a preset counting threshold, it is determined that the preset coefficient matrix adjustment condition is met.
5. The torque sensor offset acquisition method according to claim 1, characterized in that, After initializing the model parameters of the torque estimation model, the following steps are also included: Determine whether the current condition meets the torque sensor offset learning criteria; If so, then execute the torque sensor offset learning process.
6. The torque sensor offset acquisition method according to claim 1, characterized in that, The torque sensor offset learning condition is determined to be met when all of the following conditions are satisfied: The learning enable switch is currently on; The steering wheel angle, steering wheel torque, and steering wheel speed are all within their respective preset allowable ranges; as well as The steering wheel is in a state where it is not driven by the user, and the duration of this state is greater than a preset time threshold.
7. The torque sensor offset acquisition method according to claim 1, characterized in that, Determining the offset of the torque sensor based on the learned torque estimation model parameters includes the following steps: The input of the learned torque estimation model is set to zero, and the output value of the torque estimation model is obtained as the offset of the torque sensor.
8. A torque sensor offset acquisition system, characterized in that, The system for implementing the torque sensor offset acquisition method as described in any one of claims 1 to 7 includes: The parameter initialization module is used to initialize the model parameters of the torque estimation model, which takes the steering wheel angle as input and outputs the estimated torque value. The data acquisition module is used to obtain the current steering wheel angle value and the measured torque value currently collected by the torque sensor located at the steering wheel; The parameter learning module is used to obtain the estimated torque value based on the steering wheel angle value and the torque estimation model, and to learn the model parameters of the torque estimation model based on the measured torque value and the estimated torque value. The offset acquisition module is used to determine the offset of the torque sensor based on the model parameters of the learned torque estimation model.
9. The torque sensor offset acquisition system according to claim 8, characterized in that, The parameter learning module is configured to calculate the sum of squared errors based on the measured torque value and the estimated torque value, and to learn the model parameters of the torque estimation model based on the sum of squared errors using the recursive least squares method.
10. The torque sensor offset acquisition system according to claim 9, characterized in that, It also includes a precision adjustment module, used to calculate the difference between the estimated torque value and the measured torque value; and to determine whether the preset coefficient matrix adjustment conditions are met based on the difference. If so, adjust the coefficient matrix of the recursive least squares method to improve the tracking speed of the recursive least squares method.