A vehicle model parameter joint identification method considering unknown noise influence

Abstract

The application discloses a vehicle model parameter joint identification method considering unknown noise influence, relates to the technical field of new energy vehicle design, and solves the technical problem of low estimation accuracy of the total mass and the mass center position. The joint identification method is composed of a first robust embedded cubature Kalman filter and a second robust embedded cubature Kalman filter. First, the H-infinite filter is combined with the embedded cubature Kalman filter to form the first robust embedded cubature Kalman filter, and the total mass is estimated by using a vehicle longitudinal dynamics model; then, the result is input into the second robust embedded cubature Kalman filter for mass center position parameter identification. The application can realize high-precision estimation of the vehicle mass and the mass center position parameters under the condition that the statistical characteristics of noise are unknown, and can effectively improve the vehicle chassis control performance.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of new energy vehicle design, and particularly relates to a vehicle model parameter joint identification method considering unknown noise influence. BACKGROUND

[0002] Accurate vehicle parameters are of great significance to the design of vehicle motion controllers. The inertia and geometric parameters of distributed drive electric vehicles change significantly due to changes in loading conditions, so the distributed drive technology developed in recent years urgently needs more accurate and reliable online vehicle parameter estimation methods. Among all vehicle parameters, the curb weight and the position of the center of mass are two important parameters, for example, economic driving, active cruise control, anti-lock braking system control and other strategies need to know the mass of the vehicle. On the other hand, the position of the center of mass must be known when controlling the yaw moment of the vehicle, which is difficult to obtain directly from the current vehicle sensors, so the estimation of the curb weight and the position of the center of mass has always been a core technology of automobiles. However, the existing research on estimating vehicle model parameters does not fully utilize the advantages of joint estimation, and also ignores the case where the statistical characteristics of the measurement noise are unknown, which leads to the fact that these technologies cannot be applied in real vehicles in complex scenarios. SUMMARY

[0003] The present application provides a vehicle model parameter joint identification method considering unknown noise influence, which aims to consider the case where the statistical characteristics of the noise are unknown and improve the estimation accuracy of the curb weight and the position of the center of mass.

[0004] The above technical purpose of the present application is achieved by the following technical scheme:

[0005] A vehicle model parameter joint identification method considering unknown noise influence, which is realized by a first robust embedded cubature Kalman filter and a second robust embedded cubature Kalman filter, wherein the first robust embedded cubature Kalman filter and the second robust embedded cubature Kalman filter are both composed of an H-infinity filter and an embedded cubature Kalman filter, and the method comprises the following steps:

[0006] S1: obtaining relevant information in the vehicle driving process, wherein the relevant information includes longitudinal acceleration, longitudinal speed, wheel speed and driving motor torque;

[0007] According to the longitudinal speed, the air resistance and the rolling resistance are obtained;

[0008] The wheel speed and the driving motor torque are input into a tire longitudinal force calculation model for calculation to obtain the wheel longitudinal force;

[0009] S2: In the first robust embedded capacitive Kalman filter, the longitudinal dynamic model is transformed into the mass estimation state equation and measurement equation, and the air resistance, rolling resistance and wheel longitudinal force are input into the first robust embedded capacitive Kalman filter for mass estimation to obtain the mass parameters;

[0010] S3: Input the mass parameters into the theoretical calculation model of tire vertical force to obtain the theoretically calculated tire vertical force;

[0011] The tire slip ratio is calculated based on the longitudinal velocity, wheel speed, and wheel rolling radius. The tire slip ratio and wheel longitudinal force are input into a linear tire model for calculation to obtain the indirectly measured tire vertical force.

[0012] S4: Construct the centroid estimation state equation and measurement equation based on the theoretical calculation model of tire vertical force. Input the theoretically calculated tire vertical force and the indirectly measured tire vertical force into the second robust embedded capacitive Kalman filter to estimate the centroid position and obtain the centroid position.

[0013] S5: Use the current mass parameters and centroid position as inputs for the state estimation at the next moment, and repeat steps S2 to S4 until the estimation accuracy of the vehicle model parameters reaches the preset standard.

[0014] The beneficial effects of this application are as follows: The joint identification method for vehicle model parameters under the influence of unknown noise described in this application considers the unknown statistical characteristics of the measured noise, and uses a robust embedded capacitive Kalman filter to estimate the vehicle model parameters, establishes a joint estimation framework for mass and centroid position, realizes high-precision estimation of mass and centroid position, and expands the applicability of this application. Attached Figure Description

[0015] Figure 1 This is a flowchart of a method for joint identification of vehicle model parameters considering the influence of unknown noise in a specific embodiment of this application;

[0016] Figure 2 This is a schematic diagram for calculating the longitudinal force of a wheel. Detailed Implementation

[0017] The technical solution of this application will now be described in detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0018] like Figure 1As shown, the joint identification method for vehicle model parameters considering the influence of unknown noise is implemented through a first robust embedded capacitive Kalman filter and a second robust embedded capacitive Kalman filter. Both the first and second robust embedded capacitive Kalman filters are composed of an H-infinity filter and an embedded capacitive Kalman filter. First, the curb weight is estimated using the first robust embedded capacitive Kalman filter in conjunction with the vehicle's longitudinal dynamics model. Then, the result is input into the second robust embedded capacitive Kalman filter for centroid position parameter identification.

[0019] The above methods specifically include:

[0020] S1: Acquire relevant information during vehicle operation, including longitudinal acceleration, longitudinal velocity, wheel speed, and drive motor torque.

[0021] Air resistance and rolling resistance are calculated based on longitudinal velocity.

[0022] The wheel speed and drive motor torque are input into the tire longitudinal force calculation model for calculation, and the wheel longitudinal force is obtained.

[0023] Specifically, the vehicle's longitudinal acceleration and longitudinal velocity are obtained through onboard gyroscope sensors and a satellite positioning system. The wheel longitudinal force is calculated using wheel speed sensors and drive torque information from the onboard local area network communication bus; specifically, the tire longitudinal force is calculated based on wheel speed and drive motor torque.

[0024] like Figure 2 As shown, the tire longitudinal force calculation model is expressed as follows:

[0025] F xf =(T f -Jω f ) / r;

[0026] F xr =(T r -Jω r ) / r;

[0027] Among them, T f T represents the front-wheel drive torque. r Represents the rear-wheel drive torque; ω f Indicates the front wheel speed; ω r The value represents the rear wheel speed; r represents the wheel rolling radius; and J represents the wheel moment of inertia.

[0028] S2: In the first robust embedded capacitive Kalman filter, the longitudinal dynamic model is transformed into the mass estimation state equation and measurement equation, and the air resistance, rolling resistance and wheel longitudinal force are input into the first robust embedded capacitive Kalman filter for mass estimation to obtain the mass parameters.

[0029] Specifically, the longitudinal dynamic model is expressed as:

[0030]

[0031]

[0032]

[0033] Where m represents the curb weight; v x F represents the longitudinal speed of the vehicle. xf F represents the longitudinal force on the front wheel. xr F represents the longitudinal force on the rear wheel. zf This represents the theoretically calculated vertical force on the front wheel; F zr This indicates the theoretically calculated vertical force on the rear wheel; C D ρ represents the drag coefficient; A represents the vehicle's frontal area; ρ represents the air density; g represents the acceleration due to gravity; f represents the rolling resistance coefficient; a represents the distance from the front axle to the center of gravity; b represents the distance from the rear axle to the center of gravity; a x It represents longitudinal acceleration.

[0034] S3: Input the mass parameters into the theoretical calculation model of tire vertical force to obtain the theoretically calculated tire vertical force.

[0035] The tire slip ratio is calculated based on the longitudinal velocity, wheel speed, and wheel rolling radius. The tire slip ratio and wheel longitudinal force are input into a linear tire model for calculation, resulting in an indirect measurement of the tire vertical force.

[0036] The theoretical calculation model for tire vertical force is as follows:

[0037]

[0038]

[0039] The linear tire model is represented as:

[0040]

[0041]

[0042] Where i = f and r represent the front wheel and the rear wheel, respectively; The vertical force of the tire, indirectly measured by a linear tire model, is represented by T; the torque of the wheel drive motor is represented by s. i Indicates wheel slip ratio; C xi The slope of the tire slip ratio curve is represented by r. i This indicates the rolling radius of the wheel.

[0043] S4: Construct the centroid estimation state equation and measurement equation based on the theoretical calculation model of tire vertical force. Input the theoretically calculated tire vertical force and the indirectly measured tire vertical force into the second robust embedded capacitive Kalman filter to estimate the centroid position and obtain the centroid position.

[0044] Specifically, the state equation and measurement equation for mass estimation are expressed as follows:

[0045] x k =[m] T , z k =[a x ] T ;

[0046] The centroid estimation state equation and measurement equation are expressed as follows:

[0047] x k =[a] T ,

[0048] in,

[0049] Represents the state vector of the system; Indicates a known control input; Represents the observation vector of the system; function and Let x0 represent the nonlinear state function and the measurement function, respectively; the initial state x0 is uncorrelated with w and v; v and w are uncorrelated Gaussian white noise and follow a normal distribution: wN(q,Q). It represents the set of real numbers.

[0050] As a specific embodiment, the robust embedded capsular Kalman filter of the first robust embedded capsular Kalman filter and the second robust embedded capsular Kalman filter is expressed as follows:

[0051]

[0052]

[0053] Where P0 represents the initial value of the error covariance; E(.) represents the expected value.

[0054] The time update process includes:

[0055] For the error covariance matrix P k-1 / k-1 Factorize it, and express it as:

[0056] The volume point is calculated and represented as follows:

[0057] in, n represents the dimension of the state to be estimated;

[0058] The volume point propagation is then represented as:

[0059] Further predictions from the state estimate are expressed as follows:

[0060] in,

[0061] The estimation error covariance matrix is ​​then expressed as: Where Q represents the covariance matrix corresponding to the process noise.

[0062] The measurement update process includes:

[0063] For the updated error covariance matrix P k|k-1 Factorize it, and express it as:

[0064] The updated volume points are calculated and represented as follows:

[0065] The updated volume point propagation representation is then:

[0066] The updated state estimate prediction is expressed as follows:

[0067] The updated estimated error covariance matrix is ​​expressed as:

[0068] The estimated error cross-covariance matrix is ​​then expressed as:

[0069] The filter gain is expressed as: W k =P xz,k / k-1 (P zz,k / k-1 +R) -1 ;

[0070] State update is represented as:

[0071] The updated error covariance matrix is ​​expressed as:

[0072]

[0073] Where λ represents an undetermined constant; I n Let R represent the n-dimensional identity matrix, and let R represent the measurement noise covariance matrix.

[0074] S5: Use the current mass parameters and centroid position as inputs for the state estimation at the next moment, and repeat steps S2 to S4 until the estimation accuracy of the vehicle model parameters reaches the preset standard.

[0075] Those skilled in the art will understand that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. It should also be understood that terms such as those defined in general dictionaries should be understood to have the meaning consistent with their meaning in the context of the prior art, and should not be interpreted in an idealized or overly formal sense unless defined as herein.

[0076] Based on the above-described preferred embodiments of the present invention, and through the foregoing description, those skilled in the art can make various changes and modifications without departing from the inventive concept. The technical scope of this invention is not limited to the contents of the specification, but must be determined according to the scope of the claims.

Claims

1. A method for joint identification of vehicle model parameters considering the influence of unknown noise, characterized in that, This method is implemented using a first robust embedded capacitive Kalman filter and a second robust embedded capacitive Kalman filter. Both the first and second robust embedded capacitive Kalman filters are composed of an H-infinity filter and an embedded capacitive Kalman filter. The method includes: S1: Acquire relevant information during vehicle operation, including longitudinal acceleration, longitudinal velocity, wheel speed, and drive motor torque; Air resistance and rolling resistance are calculated based on longitudinal velocity. The wheel speed and drive motor torque are input into the tire longitudinal force calculation model for calculation to obtain the wheel longitudinal force. S2: In the first robust embedded capacitive Kalman filter, the longitudinal dynamic model is transformed into the mass estimation state equation and measurement equation, and the air resistance, rolling resistance and wheel longitudinal force are input into the first robust embedded capacitive Kalman filter for mass estimation to obtain the mass parameters; S3: Input the mass parameters into the theoretical calculation model of tire vertical force to obtain the theoretically calculated tire vertical force; The tire slip ratio is calculated based on the longitudinal velocity, wheel speed, and wheel rolling radius. The tire slip ratio and wheel longitudinal force are input into a linear tire model for calculation to obtain the indirectly measured tire vertical force. S4: Construct the centroid estimation state equation and measurement equation based on the theoretical calculation model of tire vertical force. Input the theoretically calculated tire vertical force and the indirectly measured tire vertical force into the second robust embedded capacitive Kalman filter to estimate the centroid position and obtain the centroid position. S5: Use the current mass parameters and centroid position as inputs for the state estimation at the next moment, and repeat steps S2 to S4 until the estimation accuracy of the vehicle model parameters reaches the preset standard.

2. The method as described in claim 1, characterized in that, The tire longitudinal force calculation model is expressed as follows: F xf =(T f -Jω f ) / r; F xr =(T r -Jω r ) / r; Among them, T f T represents the front-wheel drive torque. r Represents the rear-wheel drive torque; ω f Indicates the front wheel speed; ω r The value represents the rear wheel speed; r represents the wheel rolling radius; and J represents the wheel moment of inertia.

3. The method as described in claim 1, characterized in that, The longitudinal dynamic model is expressed as follows: Where m represents the curb weight; v x F represents the longitudinal speed of the vehicle. xf F represents the longitudinal force on the front wheel. xr F represents the longitudinal force on the rear wheel. zf This represents the theoretically calculated vertical force on the front wheel; F zr This indicates the theoretically calculated vertical force on the rear wheel; C D ρ represents the drag coefficient; A represents the vehicle's frontal area; ρ represents the air density; g represents the acceleration due to gravity; f represents the rolling resistance coefficient; a represents the distance from the front axle to the center of gravity; b represents the distance from the rear axle to the center of gravity; a x It represents longitudinal acceleration.

4. The method as described in claim 3, characterized in that, The theoretical calculation of tire vertical force is performed using a theoretical calculation model for tire vertical force, which is expressed as follows: Among them, F zf This represents the theoretically calculated vertical force on the front wheel; F zr This indicates the theoretically calculated vertical force on the rear wheel.

5. The method as described in claim 4, characterized in that, The indirect measurement of tire vertical force is calculated using a linear tire model, which is expressed as follows: Where i = f and r represent the front wheel and the rear wheel, respectively; The vertical force of the tire, indirectly measured by a linear tire model, is represented by T; the torque of the wheel drive motor is represented by s. i Indicates tire slip ratio; C xi The slope of the tire slip ratio curve is represented by r. i This indicates the rolling radius of the wheel.

6. The method as described in claim 5, characterized in that, The state equation and measurement equation for mass estimation are expressed as follows: x k =[m] T ,z k =[a x ] T ; The centroid estimation state equation and measurement equation are expressed as follows: in, Represents the state vector of the system; Indicates a known control input; Represents the system's observation vector; function f: and h: Let x0 and w represent the nonlinear state function and measurement function, respectively; the initial state x0 is uncorrelated with w and v; v and w are uncorrelated Gaussian white noise and follow a normal distribution. It represents the set of real numbers.

7. The method as described in claim 6, characterized in that, The robust embedded capacitive Kalman filter of the first robust embedded capacitive Kalman filter and the second robust embedded capacitive Kalman filter are expressed as follows: Where P0 represents the initial value of the error covariance; E(.) represents the expected value. The time update process is as follows: For the error covariance matrix P k-1 / k-1 Factorize it, and express it as: The volume point is calculated and represented as follows: in, n represents the dimension of the state to be estimated; The volume point propagation is then represented as: Further predictions from the state estimate are expressed as follows: in, The estimation error covariance matrix is ​​then expressed as: Where Q represents the covariance matrix corresponding to the process noise; The measurement update process is as follows: For the updated error covariance matrix R k|k-1 Factorize it, and express it as: The updated volume points are calculated and represented as follows: The updated volume point propagation representation is then: The updated state estimate prediction is expressed as follows: The updated estimated error covariance matrix is ​​expressed as: The estimated error cross-covariance matrix is ​​then expressed as: The filter gain is expressed as: W k =P xz,k / k-1 (P zz,k / k-1 +R) -1 ; State update is represented as: The updated error covariance matrix is ​​expressed as: Where λ represents an undetermined constant; I n Let R represent the n-dimensional identity matrix, and let R represent the measurement noise covariance matrix.

8. The method as described in claim 1, characterized in that, In step S1, the vehicle's longitudinal acceleration and longitudinal velocity are obtained through onboard sensor gyroscopes and satellite positioning systems.

9. The method as described in claim 1, characterized in that, In step S1, the longitudinal force of the wheel is calculated using the driving torque information from the wheel speed sensor and the vehicle local area network communication bus. The longitudinal force of the tire is calculated based on the wheel speed and the torque of the drive motor.

Images