CN112270039A - Nonlinear state estimation method for drive-by-wire chassis vehicles based on distributed asynchronous fusion - Google Patents

Abstract

The invention discloses a nonlinear state estimation method of a wire-controlled chassis vehicle based on distributed asynchronous fusion. The steps are as follows: 1) establishing a four-degree-of-freedom motion differential equation of the vehicle including the longitudinal, lateral, yaw and roll motions of the center of mass; 2) establishing a vehicle nonlinear state equation and an observation equation according to the vehicle four-degree-of-freedom motion differential equation; 3) iterating the state parameters in the vehicle nonlinear state equation to a nonlinear state time-delay volume Kalman fusion filter, The non-linear state fusion estimation value of the vehicle is obtained, which is used for the real-time fusion estimation of the key control variables of the drive-by-wire chassis system. The present invention effectively solves the state time delay problem that occurs when the sensor observes and describes the driving state of the vehicle due to the different sampling frequencies of different on-board sensors.

Description

Nonlinear state estimation method for drive-by-wire chassis vehicles based on distributed asynchronous fusion

technical field

The invention belongs to the field of intelligent driving environment perception, and in particular relates to a nonlinear state estimation method of a wire-controlled chassis vehicle based on distributed asynchronous fusion.

Background technique

In recent years, with the in-depth application of information technology in the field of automobiles, intelligent driving technology has been further developed and improved, and the control of automobile chassis by wire has become a major trend in the development of modern automobiles. The key to the control of the drive-by-wire chassis system technology is to accurately obtain key state variables such as yaw rate, longitudinal and lateral speed, and body roll angle that characterize the running state of the vehicle itself. These state variables are the main control variables in the vehicle-by-wire chassis control system, and are also an important basis for real-time identification of the vehicle's driving state and formulation of coordinated control rules for the drive-by-wire chassis subsystem. However, many key state variables cannot be measured directly, accurately or at low cost due to the complexity of the vehicle dynamics control process and the testing level and testing cost of on-board sensors.

Existing vehicle driving state estimation methods (such as authorized publication number CN106250591B) mainly establish motion differential equations with nonlinear characteristics such as vehicle mass center motion, yaw motion and roll motion, and then use extended Kalman filtering to indirectly calculate the vehicle state. Parameter Estimation. However, in practical applications, due to the different sampling frequencies of different on-board sensors, the state time lag occurs when the sensor observes and describes the driving state of the vehicle, which brings great influence to the accurate description of system noise and statistical characteristics of observation noise Therefore, if a conventional filter with a pre-established noise model is used, there will be inaccurate state estimation and even divergence, which makes it impossible to realize the precise control of the vehicle chassis subsystem by the wire-controlled chassis control center.

SUMMARY OF THE INVENTION

In view of the above-mentioned deficiencies of the prior art, the purpose of the present invention is to provide a nonlinear state estimation method for a chassis-by-wire vehicle based on distributed asynchronous fusion. The filtering theory and multi-sensor information asynchronous fusion technology are introduced into the nonlinear state estimation of the vehicle, and the nonlinear state time-delay volume Kalman fusion filter is designed, which effectively solves the problem of the sensor's impact on the vehicle's driving state caused by the different sampling frequencies of different on-board sensors. The state time-delay problem that occurs in the observation and description realizes the real-time fusion estimation of the key variables for active control of the drive-by-wire chassis system, and provides a more accurate signal for the active safety control of the vehicle.

For achieving the above object, the technical scheme adopted in the present invention is as follows:

A method for estimating the nonlinear state of a chassis-by-wire vehicle based on distributed asynchronous fusion of the present invention, the steps are as follows:

1) Establish a differential equation of four-degree-of-freedom motion of the vehicle including the longitudinal, lateral, yaw and roll motions of the center of mass;

2) establishing a nonlinear state equation and an observation equation of the vehicle according to the four-degree-of-freedom motion differential equation of the vehicle;

3) Iterating the state parameters in the nonlinear state equation of the vehicle to the nonlinear state time-delay volume Kalman fusion filter to obtain the vehicle nonlinear state fusion estimation value, which is used for real-time fusion estimation of the key variables of the drive-by-wire chassis system control .

Further, the differential equation of motion in the step 1) is:

where m is the vehicle mass, u is the longitudinal speed, v is the lateral speed, φ and r are the roll and yaw angular velocities, respectively; h ₀ is the distance from the center of mass to the roll axis; h _r is the height of the roll center; h _rf and h _rr are the distance from the front and rear axles to the roll axis respectively; ε is the angle between the roll axis and the longitudinal axis; δ is the front wheel rotation angle; I _b,x is the moment of inertia around the X axis, I _b,z is the moment of inertia around the Z axis, I _b,xz is the inertia product around the X and Z axes; t _f is the front track, t _r is the rear track; k _f is the cornering stiffness of the front tire, and k _r is the rear tire cornering stiffness; a is the distance from the center of mass to the front axle, b is the distance from the center of mass to the rear axle; c _f is the roll angle damping of the front suspension, _cr is the roll angle damping of the rear suspension; F _y1 is the left front wheel lateral force, F _y2 is the lateral force of the right front wheel, F _y3 is the lateral force of the left rear wheel, F _y4 is the lateral force of the right rear wheel; F _x1 is the longitudinal force of the left front wheel, F _x2 is the longitudinal force of the right front wheel, and F _x3 is the left Rear wheel longitudinal force, F _x4 right rear wheel longitudinal force.

Further, the nonlinear state equation and observation equation of the vehicle in the step 2) are:

为k+1时刻的状态估计值，x_k为线控底盘主动控制状态变量，f(·)表示系统动力学函数，过程噪声w_k是均值为零、协方差矩阵为Q_k的高斯白噪声；z_k,i是在t_k,i时刻获得的测量值，N_k个观测值是由具有不同采样频率的m个传感器在时间间隔[t_k-1,t_k]内获得，在采样时间序列内获得的N_k个观测值

是异步的，满足关系：

h_k,i表示测量函数，x_k,i表示在t_k,i时刻的状态量，测量噪声η_k,i是均值为零、协方差矩阵为R_k,i的高斯白噪声。In the formula,

is the state estimate value at time k+1, x _k is the active control state variable of the drive-by-wire chassis, f( ) represents the system dynamics function, and the process noise w _k is Gaussian white noise with zero mean and covariance matrix Q _k ; z _k,i is the measurement obtained at time t _k,i , N _k observations are obtained by m sensors with different sampling frequencies in the time interval [t _k-1 ,t _k ], at the sampling time N _k observations obtained within the sequence

is asynchronous and satisfies the relation:

h _k,i represents the measurement function, x _k,i represents the state quantity at time t _k,i , and the measurement noise η _k,i is Gaussian white noise with zero mean and covariance matrix R _k,i .

Further, the state prediction equation of the nonlinear state time-delay volume Kalman fusion filter in the step 3) is:

The prediction covariance matrix is:

定义如下：In the formula, X _k-1 ,

Defined as follows:

X _k-1 =x _k-1-j

In the formula, j,l∈[0,1,…d-1] and j≠l, s,t∈[0,1,…,d-1], d is the number of volume points.

Further, the state update equation of the medium nonlinear state time-delay volume Kalman fusion filter is:

The updated covariance matrix is:

The gain update equation is:

in,

及其误差协方差矩阵P_k的计算公式如下：Further, the distributed fusion estimation of the nonlinear state time-delay volume Kalman fusion filter at time _tk is

The calculation formula of and its error covariance matrix P _k is as follows:

P _k =[(P _k|k-1 ) ^-1 +X _k +Y _k (P _k|k-1 -L _k ) ^-1 Y _k ^T ] ^-1

In the formula,

E _k,i =IQ(t _k ,t _k,i )(P _k|k-1 ) ^-1 ,i=1,...,N _k

and

In the formula, Φ(t _k , t _{k, i+1} ) represents the system state transition matrix from t _k to t _{k, i+1} , and Φ ^T (t _k , t _{k, i+1} ) represents the state transition The transpose matrix of the matrix; M _k,i is the information matrix at time t _k,i , and I _n represents the n-order identity matrix.

Further, the key variables in the step 3) include: yaw angular velocity, longitudinal and lateral velocity, and body roll angle.

Beneficial effects of the present invention:

Based on the nonlinear dynamic model of the vehicle, the present invention introduces the volume Kalman filter theory and multi-sensor information asynchronous fusion technology into the nonlinear state estimation of the vehicle, and designs a nonlinear state time-delay volume Kalman fusion filter, which is effective It solves the state time lag problem that occurs when the sensor observes and describes the driving state of the vehicle due to the different sampling frequencies of different on-board sensors, and realizes the real-time fusion estimation of key variables for active control of the drive-by-wire chassis system, which is the active safety of the vehicle. Control provides a more precise signal.

Description of drawings

Figure 1 is a schematic diagram of the method of the present invention.

Detailed ways

In order to facilitate the understanding of those skilled in the art, the present invention will be further described below with reference to the embodiments and the accompanying drawings, and the contents mentioned in the embodiments are not intended to limit the present invention.

Referring to FIG. 1 , a method for estimating nonlinear state of a controlled-by-wire chassis vehicle based on distributed asynchronous fusion of the present invention, the steps are as follows:

In order to reflect the nonlinear state of the vehicle more comprehensively, the key state variables necessary for the active control of the drive-by-wire chassis system but difficult to obtain directly are estimated by real-time fusion. The differential equation for the yaw motion is as follows:

The force balance equation in the X direction:

where u is the longitudinal velocity, v is the lateral velocity; φ and r are the roll and yaw angular velocities, respectively; h ₀ is the distance from the center of mass to the roll axis; ε is the angle between the roll axis and the longitudinal axis; δ is the front wheel rotation angle; F _y1 is the lateral force of the left front wheel, F _y2 is the lateral force of the right front wheel; F _x1 is the longitudinal force of the left front wheel, F _x2 is the longitudinal force of the right front wheel, and F _x3 is the left rear wheel Longitudinal force, F _x4 right rear wheel longitudinal force.

The force balance equation in the Y direction:

In the formula, F _y3 is the lateral force of the left rear wheel, and F _y4 is the lateral force of the right rear wheel.

Moment equation about the Z axis:

In the formula, I _b,x is the moment of inertia around the X axis, I _b,z is the moment of inertia around the Z axis, I _b,xz is the inertia product around the X and Z axes; a is the distance from the center of mass to the front axis, and b is the center of mass Distance to the rear axle; t _f is the front track, t _r is the rear track.

Moment equation about the X axis:

In the formula, h _r is the height of the roll center; h _rf and hr are the distance from the front and rear axles to the roll axle respectively; k _f is the cornering stiffness of the front tire, k _r is the cornering stiffness of the rear tire _; c _f is the front suspension side Inclination damping, _cr is the rear suspension roll damping.

According to the estimated object, the state equation and the measurement equation are established, the nonlinear model is linearized and the initial value is assigned to perform recursive estimation, which mainly includes the prediction process and the correction process. The specific process is as follows:

Step 1): Establish vehicle nonlinear state equation and observation equation

为k+1时刻的状态估计值，x_k为线控底盘主动控制状态变量，f(·)表示系统动力学函数，过程噪声w_k是均值为零、协方差矩阵为Q_k的高斯白噪声；z_k,i是在t_k,i时刻获得的测量值，N_k个观测值是由具有不同采样频率的m个传感器在时间间隔[t_k-1,t_k]内获得，在采样时间序列内获得的N_k个观测值

是异步的，满足关系：

h_k,i表示测量函数，x_k,i表示在t_k,i时刻的状态量，测量噪声η_k,i是均值为零、协方差矩阵为R_k,i的高斯白噪声。In the formula,

is the state estimate value at time k+1, x _k is the active control state variable of the drive-by-wire chassis, f( ) represents the system dynamics function, and the process noise w _k is Gaussian white noise with zero mean and covariance matrix Q _k ; z _k,i is the measurement obtained at time t _k,i , N _k observations are obtained by m sensors with different sampling frequencies in the time interval [t _k-1 ,t _k ], at the sampling time N _k observations obtained within the sequence

is asynchronous and satisfies the relation:

h _k,i represents the measurement function, x _k,i represents the state quantity at time t _k,i , and the measurement noise η _k,i is Gaussian white noise with zero mean and covariance matrix R _k,i .

Step 2): Calculate the predicted volume point vector, and the specific calculation method is as follows:

Cholesky decomposition of the covariance matrix P _k-1|k-1 :

Volume point calculation:

In the formula, ξ _b represents the b-th column vector of the volume point set {ξ}, and the set {ξ} has a total of 2n column vectors, which are defined as follows:

Volume point spread:

表示预测容积点向量。In the formula,

represents a vector of predicted volume points.

Step 3): The state and covariance matrix one-step prediction equation is:

Step 4): Calculate the measurement volume point vector, and the specific calculation method is as follows:

Cholesky decomposition of the covariance matrix P _k|k-1 :

Volume point calculation:

Volume point propagation: Z _b,k|k-1 =h( _Xb,k|k-1 ),b=1,...,2n

一步协方差阵

一步互协方差阵

方程为：Step 5): Measurement and prediction

one-step covariance matrix

one-step cross-covariance matrix

The equation is:

Step 6): Gain matrix update:

Step 7): Covariance matrix update:

Step 8): State variable update estimation:

和相关信息矩阵M_k,i均是异步获得的，在t_k时刻的分布式融合估计

及其误差协方差矩阵P_k的计算公式如下：Step 9): Information fusion process; local information state reconstruction of each sensor

and the related information matrix M _k,i are obtained asynchronously, and the distributed fusion estimation at time t _k

The calculation formula of and its error covariance matrix P _k is as follows:

P _k =[(P _k|k-1 ) ^-1 +X _k +Y _k (P _k|k-1 -L _k ) ^-1 Y _k ^T ] ^-1

in,

and

The specific process of the above formula is:

E _k,1 =IQ(t _k ,t _k,i )(P _k|k-1 ) ^-1 ,i=1,...,N _k

通过局部传感器节点获得，具体计算方法如下：Information Matrix M _k,i and Information State Reconstruction

Obtained through local sensor nodes, the specific calculation method is as follows:

为传感器的增广测量矩阵，完成测量函数h_k,i从时刻t_k,i到时刻t_k的过渡，其计算公式如下：In the formula,

is the augmented measurement matrix of the sensor, which completes the transition of the measurement function h _k,i from time t _k,i to time t _k , and its calculation formula is as follows:

表示x_k,i和时刻t_k,i到t_k的状态

之间的协方差；

表示状态x_k,i和测量z_k,i间的互协方差矩阵；计算方式如下：In the formula,

Represents the state of x _k,i and time tk _,i to _tk

covariance between;

represents the cross-covariance matrix between states x _k,i and measurements z _k,i ; it is calculated as follows:

Volume point calculation:

In the formula, S _k,i- is obtained by Cholesky decomposition of the covariance matrix P _k,i- :

Volume point spread:

表示从时刻t_k',i'到t_k,i的传播状态

的预测观测向量，

表示从时刻t_k,i到t_k的状态。In the formula,

Represents the propagation state from time t _k',i' to t _k,i

The predicted observation vector of ,

represents the state from time t _k,i to t _k .

There are many specific application ways of the present invention, and the above are only the preferred embodiments of the present invention. It should be pointed out that for those skilled in the art, without departing from the principle of the present invention, several improvements can be made. These Improvements should also be considered as the protection scope of the present invention.

Claims (7)

1. a method for estimating the nonlinear state of a chassis-by-wire vehicle based on distributed asynchronous fusion, is characterized in that, step is as follows: 1) Establish a differential equation of four-degree-of-freedom motion of the vehicle including the longitudinal, lateral, yaw and roll motions of the center of mass; 2) establishing a nonlinear state equation and an observation equation of the vehicle according to the four-degree-of-freedom motion differential equation of the vehicle; 3) Iterating the state parameters in the nonlinear state equation of the vehicle to the nonlinear state time-delay volume Kalman fusion filter to obtain the vehicle nonlinear state fusion estimation value, which is used for real-time fusion estimation of the key variables of the drive-by-wire chassis system control . 2. the non-linear state estimation method of the chassis-by-wire vehicle based on distributed asynchronous fusion according to claim 1, is characterized in that, the differential equation of motion in described step 1) is:

where m is the vehicle mass, u is the longitudinal speed, v is the lateral speed, φ and r are the roll and yaw angular velocities, respectively; h ₀ is the distance from the center of mass to the roll axis; h _r is the height of the roll center; h _rf and h _rr are the distance from the front and rear axles to the roll axis respectively; ε is the angle between the roll axis and the longitudinal axis; δ is the front wheel rotation angle; I _b,x is the moment of inertia around the X axis, I _b,z is the moment of inertia around the Z axis, I _b,xz is the inertia product around the X and Z axes; t _f is the front track, t _r is the rear track; k _f is the cornering stiffness of the front tire, and k _r is the rear tire cornering stiffness; a is the distance from the center of mass to the front axle, b is the distance from the center of mass to the rear axle; c _f is the roll angle damping of the front suspension, _cr is the roll angle damping of the rear suspension; F _y1 is the left front wheel lateral force, F _y2 is the lateral force of the right front wheel, F _y3 is the lateral force of the left rear wheel, F _y4 is the lateral force of the right rear wheel; F _x1 is the longitudinal force of the left front wheel, F _x2 is the longitudinal force of the right front wheel, and F _x3 is the left Rear wheel longitudinal force, F _x4 right rear wheel longitudinal force. 3. the non-linear state estimation method of the chassis-by-wire vehicle based on distributed asynchronous fusion according to claim 1, is characterized in that, the non-linear state equation and observation equation of vehicle in described step 2) are:

In the formula,

is the state estimate value at time k+1, x _k is the active control state variable of the drive-by-wire chassis, f( ) represents the system dynamics function, and the process noise w _k is Gaussian white noise with zero mean and covariance matrix Q _k ; z _k,i is the measurement obtained at time t _k,i , N _k observations are obtained by m sensors with different sampling frequencies in the time interval [t _k-1 ,t _k ], at the sampling time N _k observations obtained within the sequence

is asynchronous and satisfies the relation:

h _k,i represents the measurement function, x _k,i represents the state quantity at time t _k,i , and the measurement noise η _k,i is Gaussian white noise with zero mean and covariance matrix R _k,i . 4. the non-linear state estimation method of the chassis-by-wire vehicle based on distributed asynchronous fusion according to claim 1, is characterized in that, the state prediction equation of nonlinear state time-delay volume Kalman fusion filter in described step 3) for:

The prediction covariance matrix is:

In the formula, X _k-1 ,

Defined as follows: X _k-1 =x _k-1-j

In the formula, j,l∈[0,1,…d-1] and j≠l, s,t∈[0,1,…,d-1], d is the number of volume points. 5. the non-linear state estimation method of the chassis-by-wire vehicle based on distributed asynchronous fusion according to claim 1, is characterized in that, the state update equation of described medium nonlinear state time-delay volume Kalman fusion filter is:

The updated covariance matrix is:

The gain update equation is:

in,

6 . The nonlinear state estimation method for a controlled chassis vehicle based on distributed asynchronous fusion according to claim 1 , wherein the distributed fusion of the nonlinear state time-delay volume Kalman fusion filter at time t _k estimate

The calculation formula of and its error covariance matrix P _k is as follows:

In the formula,

E _k,i =IQ(t _k ,t _k,i )(P _k|k-1 ) ^-1 ,i=1,...,N _k and

In the formula, Φ(t _k , t _{k, i+1} ) represents the system state transition matrix from t _k to t _{k, i+1} , and Φ ^T (t _k , t _{k, i+1} ) represents the state transition The transpose matrix of the matrix; M _k,i is the information matrix at time t _k,i , and I _n represents the n-order identity matrix. 7. The method for estimating the nonlinear state of a chassis-by-wire vehicle based on distributed asynchronous fusion according to claim 1, wherein the key variables in the step 3) include: yaw angular velocity, longitudinal and lateral velocity, body side inclination.

Images