A Steering Return Control Method Based on Driving Style Identification

Abstract

The present invention belongs to the field of automobile technology, and in particular, is a steering return control method based on driving style identification, which is composed of a driving style identification module, a return speed selection module, and a steering return control module. The steering wheel angle peak value, steering wheel speed peak value, vehicle speed at the corresponding moment, and vehicle yaw rate peak value when the driver is steering are obtained through sensors; the data when the driver is steering are clustered by fuzzy C-means clustering and BP neural network methods to obtain the driver's driving style type; the return speed selection module selects the corresponding return speed according to different driving styles; the steering return control module returns the steering wheel at the desired speed through sliding mode control. The steering return control method proposed in the present invention can change the return speed when returning according to the driver's driving style, so that the vehicle's steering return is more in line with the driver's psychological expectations.

Description

Technical Field

[0001] The present invention relates to the field of automobile technology, and in particular to a steering return control method based on driving style identification. Background Art

[0002] Electric power steering (EPS) is widely used in modern cars due to its higher energy efficiency and response speed. Good steering return performance can ensure that the vehicle quickly returns to a straight-line state after turning. For car steering movements, high-quality steering return performance can improve vehicle stability and driving comfort.

[0003] Invention patent CN117864237A discloses a return control method for an electric power steering system based on return speed control, which is composed of a hands-off return state judgment module, a return time determination module, a return speed determination module, and a return compensation torque determination module. By designing a return time determination module, the return time is determined according to the steering wheel angle and vehicle speed at the moment of hands-off, which better meets the driver's expectations. By correcting the return speed, the return performance under various working conditions is more in line with the driver's expectations, but the invention does not take into account the different expected return speeds of drivers with different driving styles.

[0004] Invention patent CN110723200A discloses a steering return and intermediate position control system and control method thereof, which determines whether the vehicle is in a steering return or intermediate position control condition through a return judgment module of a steering integrated controller, and makes the estimated value of the center of mass slip angle approach the expected center of mass slip angle through a time-varying sliding mode control module, thereby calculating the return compensation torque during the return condition. This invention can improve the stability and accuracy of the distributed drive vehicle differential power steering return and the control accuracy in the middle position of the steering wheel, but different drivers have different driving styles and different requirements for steering return speed.

[0005] The above invention adopts different return-to-center control methods to make the return-to-center performance under various working conditions more in line with the driver's expectations, but does not consider the impact of the driver's personalized driving style on the return-to-center speed. The driver's steering feel is a relatively important part of the driving experience. Applying the driving style to the personalized steering return of the car can improve the driver's return-to-center feel to a certain extent and improve the driver's driving satisfaction. Aggressive drivers prefer a faster steering return speed to enhance the sense of control; while conservative drivers may prefer a slower steering return speed to ensure driving stability. Therefore, in order to further improve the driver's steering return feel and driving experience, it is necessary to identify the driver's driving style and flexibly adjust the return feel according to their own preferences. Summary of the invention

[0006] 1. Technical issues to be resolved

[0007] In view of the deficiencies in the prior art, the present invention provides a steering return control method based on driving style identification, which solves the problems raised in the above background technology.

[0008] (II) Technical solution

[0009] In order to achieve the above-mentioned purpose, the present invention specifically adopts the following technical solutions:

[0010] A steering return control method based on driving style identification includes a driving style identification module, a return speed selection module and a steering return control module, and also includes a sensor, through which four types of data when the driver is steering are obtained: steering wheel angle peak value , Steering wheel speed peak , the vehicle speed at the corresponding time and the peak yaw rate of the vehicle The steering return control method comprises the following steps:

[0011] Step 1: First, obtain sufficient driving data that can reflect different driving styles, where each set of driving data contains the four characteristic quantities that characterize driving styles mentioned above, namely, the peak steering wheel angle when the driver turns , Steering wheel speed peak , the vehicle speed at the corresponding time and the peak yaw rate of the vehicle ;

[0012] Step 2: Use the fuzzy C-means clustering method to cluster the above data. The specific steps are as follows:

[0013] 201. First complete the number of clusters , set the number of clusters to 3, divide the driving style into three categories: aggressive, normal, and conservative, and complete the fuzzy index , convergence accuracy and the maximum number of iterations The settings of , and initialize the membership matrix ;

[0014] ;

[0015] If yes samples, specify the number of clusters as , then the membership matrix should be The matrix of Indicates Sample pair Cluster Centers The membership degree of The samples belong to The probability of the group, the membership obviously satisfies the following formula:

[0016] ;

[0017] 202. Calculate cluster centers ;

[0018] ;

[0019] in represents the total number of samples, Indicates Sample values, ;

[0020] 203. Update the membership matrix ;

[0021] ;

[0022] in It can be any norm that measures the similarity (distance) of data. The commonly used one is the Euclidean norm (also known as the Euclidean norm, L2 norm and Euclidean distance):

[0023] ;

[0024] Check stop conditions;

[0025] Comparison of objective functions and ,if , then stop the iteration, otherwise , return to step 2, and the objective function is set as follows:

[0026] ;

[0027] The constraints are:

[0028] ;

[0029] After clustering is completed, the cluster centers corresponding to the three driving styles will be obtained. , And the degree of membership of each group of data relative to the three categories ,in They represent four characteristic quantities respectively. Subscripts 1, 2, and 3 correspond to the aggressive, normal, and conservative driving styles respectively. Respectively represent The degree of membership of the group data relative to the radical type, the ordinary type, and the conservative type;

[0030] Step 3: Then the cluster data is trained in the BP neural network, where the input is the corresponding values ​​of the four feature quantities, namely the peak value of the steering wheel angle when the driver turns , Steering wheel speed peak , the vehicle speed at the corresponding time and the peak yaw rate of the vehicle , the output is the corresponding membership vector, which represents the membership of the group of data to each driving style;

[0031] Step 4: After clustering, we get the cluster centers corresponding to the three driving styles. , , according to the obtained cluster center, the numerical ratio of each characteristic quantity of the aggressive type and the conservative type relative to the ordinary type is calculated in turn, and each ratio value is corrected accordingly and added to obtain the corresponding force curve correction coefficient of the driving style relative to the ordinary type. According to the coefficient, the ordinary type return speed is corrected to obtain the other two return speeds;

[0032] ;

[0033] ;

[0034] in are the correction coefficients of radical and conservative types relative to ordinary types, They are the proportional correction coefficients corresponding to the four characteristic quantities of the radical type and the conservative type, and the above proportional correction coefficients are calibrated through simulation and experiment;

[0035] The return speed selection module receives the driving style membership vector output by the driving style recognition module Get the membership degree of the current driving behavior relative to the three driving styles;

[0036] Step 5: According to the desired steering wheel angle obtained by the return speed selection module, the desired return torque is obtained through sliding mode control. The steps are as follows:

[0037] 501. Design the sliding surface, the system equation is as follows:

[0038] ;

[0039] The designed sliding surface is as follows:

[0040] ;

[0041] 502. Design a sliding mode control law. The sliding mode control law includes two parts: equivalent control Teq and switching control Tsw. The equivalent control makes the derivative on the sliding surface zero. Let , then:

[0042] ;

[0043] Substitute T to get:

[0044] ;

[0045] Switching control makes the system state approach the sliding surface and designs the saturation function The form is as follows:

[0046] ;

[0047] Where H is the saturation function density, and the switching control Tsw is as follows:

[0048] ;

[0049] Among them, ks is the control gain, and the overall control law is:

[0050] ;

[0051] 503. Use Lyapunov function to judge the stability of the system, let V be expressed as follows:

[0052] ;

[0053] Substituting into the above formula, we can get , the control rate can be stable;

[0054] Through the above five steps, driver style recognition, return speed selection and steering return control can be achieved.

[0055] Furthermore, the return speed selection module selects corresponding return speeds according to different driving styles; and the steering return control module returns the steering wheel to the desired speed through sliding mode control.

[0056] Furthermore, in step 1, clustering is performed on the data of the driver turning by fuzzy C-means clustering and BP neural network method to obtain the driving style type of the driver.

[0057] Furthermore, the membership degree of each driving style in step 4 The corresponding correction factor The product is obtained by multiplying the weight of the return speed, and the return speeds of the three driving styles are combined according to the obtained weight coefficient to obtain the final return speed.

[0058] Furthermore, in step five, a sliding surface is introduced and a control law is designed so that the state of the system can slide according to a specified sliding mode trajectory, thereby achieving control of the system and allowing the steering wheel to return to the desired return speed.

[0059] Furthermore, the driver's driving style is classified through driving style identification, and different steering return speeds are selected according to different types of driving styles. Then, the steering wheel can be returned at the desired return speed through sliding mode control.

[0060] (III) Beneficial effects

[0061] Compared with the prior art, the present invention provides a steering return control method based on driving style identification, which has the following beneficial effects:

[0062] 1. The steering return control method proposed in the present invention can change the return speed during return according to the driver's driving style, so that the vehicle's steering return is more in line with the driver's psychological expectations.

[0063] 2. The present invention performs clustering processing on the data of the driver's steering by using fuzzy C-means clustering and BP neural network method, and divides the driving style into three types: aggressive, normal and conservative.

[0064] 3. The present invention realizes control of the return torque through sliding mode control, so that the steering wheel can be returned at a desired return speed. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 It is a schematic diagram of the system structure of the present invention;

[0066] Figure 2 This is a schematic diagram of driving style clustering of the present invention;

[0067] Figure 3 It is a schematic diagram of the return speed under different driving styles of the present invention. DETAILED DESCRIPTION

[0068] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0069] Example

[0070] like Figure 1-3 As shown, a steering return control method based on driving style identification proposed in an embodiment of the present invention is composed of a driving style identification module, a return speed selection module and a steering return control module. The sensor obtains the peak steering wheel angle when the driver turns. , Steering wheel speed peak , the vehicle speed at the corresponding time and the peak yaw rate of the vehicle ; The driver's steering data is clustered through fuzzy C-means clustering and BP neural network methods to obtain the driver's driving style type; the return speed selection module selects the corresponding return speed according to different driving styles; the steering return control module uses sliding mode control to return the steering wheel at the desired speed.

[0071] Step 1: First, obtain sufficient driving data that can reflect different driving styles, where each set of driving data contains the four characteristic quantities that characterize driving styles mentioned above, namely, the peak steering wheel angle when the driver turns , Steering wheel speed peak , the vehicle speed at the corresponding time and the peak yaw rate of the vehicle .

[0072] Step 2: Use the fuzzy C-means clustering method to cluster the above data. The specific steps are as follows:

[0073] 201. First complete the number of clusters , set the number of clusters to 3, divide the driving style into three categories: aggressive, normal, and conservative, and complete the fuzzy index , convergence accuracy and the maximum number of iterations At the same time, the membership matrix is ​​initialized .

[0074] ;

[0075] If yes samples, specify the number of clusters as , then the membership matrix should be The matrix of Indicates Sample pair Cluster Centers The membership degree of The samples belong to The probability of a group. The membership degree obviously satisfies the following formula:

[0076] ;

[0077] 202. Calculate cluster centers ;

[0078] ;

[0079] in represents the total number of samples, Indicates Sample values, .

[0080] 203. Update the membership matrix ;

[0081] ;

[0082] in It can be any norm that measures the similarity (distance) of data. The commonly used one is the Euclidean norm (also known as the Euclidean norm, L2 norm and Euclidean distance):

[0083] ;

[0084] Check stop conditions;

[0085] Comparison of objective functions and ,if , then stop the iteration. Otherwise , return to step 2. The objective function is set as follows:

[0086] ;

[0087] The constraints are:

[0088] ;

[0089] After clustering is completed, the cluster centers corresponding to the three driving styles will be obtained. , And the degree of membership of each group of data relative to the three categories .in They represent four characteristic quantities respectively. Subscripts 1, 2, and 3 correspond to the aggressive, normal, and conservative driving styles respectively. Respectively represent The degree of membership of the group data relative to the radical type, the ordinary type and the conservative type.

[0090] The main function of the driving style identification module is to cluster the four types of data when the driver turns the above through the fuzzy C-means clustering method, and divide the driving style into three types: aggressive, normal and conservative, and set the number of clusters to 3. After the clustering is completed, the cluster centers corresponding to the three driving styles are obtained. , And the degree of membership of each group of data relative to the three categories .in, They represent four characteristic quantities respectively. Subscripts 1, 2, and 3 correspond to the aggressive, normal, and conservative driving styles respectively. Respectively represent The membership of the group data relative to the radical type, the normal type and the conservative type. The clustered data is then trained in the BP neural network, where the input is the corresponding values ​​of the four feature quantities, namely the peak value of the steering wheel angle when the driver turns , Steering wheel speed peak , the vehicle speed at the corresponding time and the peak yaw rate of the vehicle , the output is the corresponding membership vector, which represents the membership of this group of data to each driving style.

[0091] Step 3: Then the cluster data is trained in the BP neural network, where the input is the corresponding values ​​of the four feature quantities, namely the peak value of the steering wheel angle when the driver turns , Steering wheel speed peak , the vehicle speed at the corresponding time and the peak yaw rate of the vehicle , the output is the corresponding membership vector, which represents the membership of this group of data to each driving style.

[0092] Step 4: After clustering, we get the cluster centers corresponding to the three driving styles. , According to the obtained cluster centers, the numerical ratio of each characteristic quantity of the aggressive type and the conservative type relative to the ordinary type is calculated in turn, and each ratio value is corrected accordingly and added up to obtain the corresponding driving style force curve correction coefficient relative to the ordinary type. According to the coefficient, the ordinary type return speed is corrected to obtain the other two return speeds.

[0093] ;

[0094] ;

[0095] in are the correction coefficients of radical and conservative types relative to ordinary types, are the proportional correction coefficients corresponding to the four characteristic quantities of the radical type and the conservative type respectively. The above proportional correction coefficients are calibrated through simulation and experiment.

[0096] The return speed selection module receives the driving style membership vector output by the driving style recognition module Get the membership of the current driving behavior to the three driving styles. The membership of each driving style The corresponding correction factor The product is obtained by multiplying the weight of the return speed, and the return speeds of the three driving styles are combined according to the obtained weight coefficient to obtain the final return speed.

[0097] Step 5: According to the desired steering wheel angle obtained by the return speed selection module, the desired return torque is obtained through sliding mode control. The steps are as follows:

[0098] 501. Design the sliding surface, the system equation is as follows:

[0099] ;

[0100] The designed sliding surface is as follows:

[0101] ;

[0102] 502. Design a sliding mode control law, which includes two parts: equivalent control Teq and switching control Tsw. The equivalent control makes the derivative on the sliding surface zero, so , then:

[0103] ;

[0104] Substitute T to get:

[0105] ;

[0106] Switching control makes the system state approach the sliding surface and designs the saturation function The form is as follows:

[0107] ;

[0108] Where H is the saturation function density. The switching control Tsw is as follows:

[0109] ;

[0110] Where ks is the control gain. The overall control law is:

[0111] ;

[0112] 503. Use Lyapunov function to judge the stability of the system, let V be expressed as follows:

[0113] ;

[0114] Substituting into the above formula, we can get , the control rate can reach stability.

[0115] The main function of the steering return control module is to control the return torque through sliding mode control according to the desired steering wheel angle obtained by the return speed selection module. By introducing a sliding surface and designing the control law, the state of the system can slide according to the specified sliding mode trajectory, thereby realizing the control of the system and making the steering wheel return at the desired return speed.

[0116] The driver's driving style is classified through driving style identification, and different steering return speeds are selected according to different types of driving styles. Then, sliding mode control is used to make the steering wheel return at the desired return speed.

[0117] Through the above five steps, driver style recognition, return speed selection and steering return control can be achieved.

[0118] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A steering return control method based on driving style identification, comprising a driving style identification module, a return speed selection module and a steering return control module, and also comprising a sensor, through which four types of data when the driver is steering are obtained: steering wheel angle peak value , Steering wheel speed peak , the vehicle speed at the corresponding time and the peak yaw rate of the vehicle , characterized in that: The following steps are involved: Step 1: First, obtain sufficient driving data that can reflect different driving styles, where each set of driving data contains the four characteristic quantities that characterize driving styles mentioned above, namely, the peak steering wheel angle when the driver turns , Steering wheel speed peak , the vehicle speed at the corresponding time and the peak yaw rate of the vehicle ; Step 2: Use the fuzzy C-means clustering method to cluster the above data. The specific steps are as follows:

201. First complete the number of clusters , set the number of clusters to 3, divide the driving style into three categories: aggressive, normal, and conservative, and complete the fuzzy index , convergence accuracy And the maximum number of iterations The settings of , and initialize the membership matrix ; ； If yes samples, specify the number of clusters as , then the membership matrix should be The matrix of Indicates Sample pair Cluster Centers The membership degree of The samples belong to The probability of the group, the membership obviously satisfies the following formula: ； 202. Calculate cluster centers : ； in represents the total number of samples, Indicates Sample values, ; 203. Update the membership matrix : ； in is an arbitrary norm that measures the similarity of data: ； Check the stop condition: Comparison of objective functions and ,if , then stop the iteration, otherwise , return to step 202, and set the objective function as follows: ； The constraints are: ； After clustering is completed, the cluster centers corresponding to the three driving styles will be obtained. , And the degree of membership of each group of data relative to the three categories ,in They represent four characteristic quantities respectively. Subscripts 1, 2, and 3 correspond to the aggressive, normal, and conservative driving styles respectively. Respectively represent The degree of membership of the group data relative to the radical type, the ordinary type, and the conservative type; Step 3: Then the cluster data is trained in the BP neural network, where the input is the corresponding values ​​of the four feature quantities, namely the peak value of the steering wheel angle when the driver turns , Steering wheel speed peak , the vehicle speed at the corresponding time and the peak yaw rate of the vehicle , the output is the corresponding membership vector, which represents the membership of the group of data to each driving style; Step 4: After clustering, we get the cluster centers corresponding to the three driving styles. , , according to the obtained cluster center, the numerical ratio of each characteristic quantity of the aggressive type and the conservative type relative to the ordinary type is calculated in turn, and each ratio value is corrected accordingly and added to obtain the corresponding force curve correction coefficient of the driving style relative to the ordinary type. According to the coefficient, the ordinary type return speed is corrected to obtain the other two return speeds; ； ； in are the correction coefficients of radical and conservative types relative to ordinary types, They are the proportional correction coefficients corresponding to the four characteristic quantities of the radical type and the conservative type, and the above proportional correction coefficients are calibrated through simulation and experiment; The return speed selection module receives the driving style membership vector output by the driving style recognition module Get the membership of the current driving behavior relative to the three driving styles, and the membership of each driving style The corresponding correction factor The weight of the return speed is obtained by multiplying the product, and the return speeds of the three driving styles are combined according to the obtained weight coefficients to obtain the final return speed; Step 5: According to the desired steering wheel angle obtained by the return speed selection module, the desired return torque is obtained through sliding mode control. The steps are as follows:

501. Design the sliding surface, the system equation is as follows: ； The designed sliding surface is as follows: ； 502. Design a sliding mode control law. The sliding mode control law includes two parts: equivalent control Teq and switching control Tsw. The equivalent control makes the derivative on the sliding surface zero. Let , then: ； Substitute T to get: ； Switching control makes the system state approach the sliding surface and designs the saturation function The form is as follows: ； Where H is the saturation function density, and the switching control Tsw is as follows: ； Among them, ks is the control gain, and the overall control law is: ； 503. Use Lyapunov function to judge the stability of the system, let V be expressed as follows: ； Substituting into the above formula, we can get , the control rate can be stable; Through the above five steps, driver style recognition, return speed selection and steering return control can be achieved.

Images