Disclosure of Invention
The invention provides a real-time estimation method of the mass center slip angle and the tire lateral force of a vehicle for a distributed driving electric vehicle, and the estimation result has great significance for an active safety control system of the electric vehicle.
The technical scheme adopted by the invention for solving the technical problems is as follows:
the method for estimating the centroid slip angle and the tire lateral force of the distributed drive electric automobile comprises the following steps of:
step one, considering the complexity of the running condition of the vehicle and the fact that the distributed driving electric vehicle has higher degree of freedom, an eight-degree-of-freedom vehicle dynamic equation is established, wherein the eight-degree-of-freedom vehicle dynamic equation comprises longitudinal motion, transverse motion, yaw motion, roll motion and motion of four tires;
selecting a linear tire model and a nonlinear Dugoff tire model as a model set of an interactive multi-model algorithm;
estimating the mass center and the side deflection angle of the vehicle and the tire side force based on an interactive multi-model algorithm-volume Kalman filtering, wherein the algorithm process comprises the following steps: input interaction, cubature Kalman filtering, model updating probability and output interaction.
Considering longitudinal motion, transverse motion, yaw motion, roll motion and motion of four tires of the vehicle under complex working conditions in the step one, and establishing a vehicle dynamic model;
the eight-degree-of-freedom vehicle dynamics equation is established as follows:
wherein, the expressions (1) and (2) are respectively longitudinal and transverse motion equations of the vehicle body of the vehicle, the expressions (3) and (4) are respectively longitudinal and transverse motion equations of the whole vehicle, and the expressions (5) and (6) are respectively transverse and side-rolling motion equations of the whole vehicle; m, m s Respectively representing the total mass of the vehicle and the sprung mass of the vehicle; i is zz 、I xz 、I xxs 、I xzs Respectively representing the yaw moment of the vehicle, the product of the moment of inertia of the mass of the vehicle around the x and z axes, the yaw moment of the sprung mass of the vehicle and the product of the moment of inertia of the sprung mass of the vehicle around the x and z axes; v x 、V y 、ω z 、ω x Respectively representing a vehicle longitudinal speed, a vehicle lateral speed, a vehicle yaw rate and a vehicle yaw rate; beta, delta and phi respectively represent a vehicle mass center slip angle, a front wheel rotating angle and a vehicle slip angle; f x 、F y Respectively representing the longitudinal force and the lateral force of four wheels of the vehicle; subscripts i = f, r respectively denote the vehicle front axle and the vehicle rear axle, subscripts j = q, p respectively denote the vehicle left wheel and the vehicle right wheel, whereby subscripts fq, fp, rq, rp respectively denote the vehicle left front wheel, right front wheel, left rear wheelAnd a right rear wheel; h represents the vertical distance from the center of mass of the sprung part of the vehicle to the roll axis; m z 、M x Respectively representing yaw moment and roll moment; b is f 、B r Respectively showing the front and rear rail widths of the vehicle; l is f 、L r Respectively representing the front and rear wheel base of the vehicle; g represents the gravitational acceleration; the upper symbol "·" represents the differentiation of the indicated quantity.
(II) establishing an expression of each tire vertical load in the second step as follows:
wherein, F z Representing vehicle vertical forces; subscripts i = f, r respectively denote a vehicle front axle and a vehicle rear axle, subscripts j = q, p respectively denote a vehicle left wheel and a vehicle right wheel, whereby subscripts fq, fp, rq, rp respectively denote a vehicle left front wheel, a right front wheel, a left rear wheel, and a right rear wheel; m, m s 、m uf 、m ur Respectively representing the total mass of the vehicle, the sprung mass of the vehicle, and the front and rear unsprung masses of the vehicle; c φf 、C φr Respectively representing the front roll rigidity and the rear roll rigidity of the vehicle; k is φf 、K φr Respectively representing the front roll damping coefficient and the rear roll damping coefficient of the vehicle; v x 、V y 、ω z 、ω x Respectively representing a vehicle longitudinal speed, a vehicle lateral speed, a vehicle yaw rate and a vehicle yaw rate; g represents the gravitational acceleration; h is a total of cg 、h uf 、h ur Respectively representing vehiclesThe height of the center of mass and the offset centers of the front wheel and the rear wheel from the ground; respectively representing the front and rear roll center heights of the vehicle; b is f 、B r Respectively showing the front and rear rail widths of the vehicle; l is f 、L r Respectively representing the front and rear wheel base of the vehicle; phi represents a roll angle; the upper symbol "·" denotes the differentiation of the quantity indicated.
The tire slip ratio expression is:
wherein λ represents a tire slip ratio; v x 、ω ij Respectively representing the vehicle longitudinal speed and the tire yaw rate; subscripts i = f, r respectively denote a vehicle front axle and a vehicle rear axle, subscripts j = q, p respectively denote a vehicle left wheel and a vehicle right wheel, whereby subscripts fq, fp, rq, rp respectively denote a vehicle left front wheel, a right front wheel, a left rear wheel, and a right rear wheel; r e Is the effective radius of the tire.
(III) selecting a linear tire model and a nonlinear Dugoff tire model as a model set of the interactive multi-model algorithm;
the linear tire model is as follows:
the linear tire model slip angle expression is:
wherein α and δ respectively represent a tire slip angle and a front wheel rotation angle; v x 、V y 、ω z Respectively representing a vehicle longitudinal speed, a vehicle transverse speed and a vehicle yaw rate; b is f 、B r Respectively showing the front and rear rail widths of the vehicle; l is f 、L r Respectively representing the front and rear wheel base of the vehicle; phi represents a roll angle; epsilon f 、ε r Respectively representing the tilting coefficients of the front shaft side and the rear shaft side; subscripts i = f, r respectively denote the vehicle front axle and the vehicle rear axle, subscripts j = q, p respectively denote the vehicle left wheel and the vehicle right wheel, whereby subscripts fq, fp, rq, rp respectively are tabulatedShowing the left front wheel, the right front wheel, the left rear wheel and the right rear wheel of the vehicle.
The linear tire longitudinal and lateral force expressions are:
F yij =-C yij α ij
F xij =k μ λ ij F zij
wherein, F x 、F y Respectively representing the longitudinal force and the transverse force of four wheels of the vehicle; alpha and lambda respectively represent a tire slip angle and a slip ratio; c y 、k μ Indicating the cornering stiffness and the linear region mu-lambda of the tyre ij An image slope; subscripts i = f, r respectively denote a vehicle front axle and a vehicle rear axle, subscripts j = q, p respectively denote a vehicle left wheel and a vehicle right wheel, whereby subscripts fq, fp, rq, rp respectively denote a vehicle left front wheel, a right front wheel, a left rear wheel, and a right rear wheel.
The nonlinear Dugoff tire model is as follows:
the nonlinear tire model slip angle expression is as follows:
wherein α and δ represent a tire slip angle and a front wheel rotation angle; v x 、V y 、ω z Respectively representing a vehicle longitudinal speed, a vehicle transverse speed and a vehicle yaw rate; b is f 、B r Respectively showing the width of the front and the rear vehicle rails of the vehicle; l is a radical of an alcohol f 、L r Respectively representing the front and rear wheel base of the vehicle; phi represents a roll angle; subscripts i = f, r respectively denote a vehicle front axle and a vehicle rear axle, subscripts j = q, p respectively denote a vehicle left wheel and a vehicle right wheel, whereby subscripts fq, fp, rq, rp respectively denote a vehicle left front wheel, a right front wheel, a left rear wheel, and a right rear wheel.
The non-linear Dugoff tire longitudinal and lateral force expressions are:
wherein, F x 、F y Respectively representing the longitudinal force and the transverse force of four wheels of the vehicle; alpha and lambda represent the tire slip angle and the slip ratio; c y Representing the tire cornering stiffness; s is the substitution amount of the function; subscript i = f, r represents the vehicle front and rear axles, respectively, subscript j = q, p represents the vehicle left and right wheels, respectively, whereby subscripts fq, fp, rq, rp represent the vehicle left and right front, rear, and rear wheels, respectively.
(IV) in the third step, estimating the vehicle mass center slip angle and the tire lateral force based on an interactive multi-model algorithm-volume Kalman filtering, wherein the input interactive process comprises the following steps:
the model probability mu of the model j in the previous step
j (k-1) and State estimation
Gets a mixed estimate->
Assuming the linear tire model and the Dugoff nonlinear tire model are models i, j =1,2, respectively, the transfer matrix is therefore:
wherein P is a transition matrix, P in the transition matrix ij Probability of motion transfer, transfer moment, from model i to model j with element as targetThe subscript r = j for the elements in the array P, r being the number of model set models.
The prediction model is derived from the expression of model j:
wherein, mu
j (k-1) represents the model probability of model j at time k-1 of the previous step,
the prediction probability of the model j is represented,
the summation operation for model i from 1 to r.
The expression of the mixing probability from model i to model j is:
wherein, mu ij (k-1; k-1) denotes the probability of mixture from model i to model j.
The hybrid state estimate is derived from the expression of model j:
wherein,
is a state estimate of the target, based on the status of the target>
Is the hybrid state estimate for model j.
Model probability mu of all filters in the last step
j (k-1) and State estimation
Get covariance >>
The hybrid covariance estimate is derived from model j as:
where T represents the transpose of the matrix,
representing the initial state of the covariance of model j, device for combining or screening>
Representing the covariance of model i.
(V) establishing a distributed driving electric automobile model in the third step:
establishing a state equation and an observation equation of a mass center slip angle and a tire lateral force of the distributed driving electric automobile:
wherein,
is the first derivative of the state variable; f (-) is a vehicle model state equation; h (-) is the vehicle model observation equation; x (t) is a state variable; u (t) is an input variable; z (t) is an observed variable; w (t) and v (t) are zero-mean, uncorrelated white noise; />
x(t)=(β,F yfq ,F yfp ,F yrq ,F yrp ) T
u(t)=(δ,F xfq ,F xfp ,F xrq ,F xrp ,ω fq ,ω fp ,ω rq ,ω rp ) T
z(t)=(a x ,a y ,ω z ,ω x ) T
Wherein, F x 、F y Respectively representing the longitudinal force and the transverse force of four wheels of the vehicle; alpha, delta and beta respectively represent a tire slip angle, a front wheel corner and a centroid slip angle; omega z 、ω x 、ω ij Respectively representing the yaw angular velocity, the vehicle roll angle velocity and the wheel rotating speed of the vehicle; a is x 、a y Respectively representing the longitudinal acceleration and the lateral acceleration of the vehicle; subscripts i = f, r respectively denote a vehicle front axle and a vehicle rear axle, subscripts j = q, p respectively denote a vehicle left wheel and a vehicle right wheel, whereby subscripts fq, fp, rq, rp respectively denote a vehicle left front wheel, a right front wheel, a left rear wheel, and a right rear wheel.
Establishing a vehicle model state equation f (-) and an observation equation h (-) by:
the functional expression f (-) is:
the functional expression h (-) is
Wherein f is 1 -f 5 Respectively representing the state equations of the vehicle model; h is 1 -h 4 Respectively representing vehicle model observation equations; f x 、F y Respectively representing the longitudinal force and the transverse force of four wheels of the vehicle; alpha, delta and beta respectively represent a tire slip angle, a front wheel rotation angle and a mass center slip angle; λ represents a tire slip ratio; subscripts i = f, r respectively denote a vehicle front axle and a vehicle rear axle, subscripts j = q, p respectively denote a vehicle left wheel and a vehicle right wheel, whereby subscripts fq, fp, rq, rp respectively denote a vehicle left front wheel, a right front wheel, a left rear wheel, and a right rear wheel; m, m s Respectively representing the total mass of the vehicle and the sprung mass of the vehicle; omega z 、ω x Respectively representing the vehicle yaw angular velocity and the vehicle side slip angular velocity; g represents the gravitational acceleration; phi represents a roll angle; h denotes a vehicleThe center of mass of the sprung portion is at a perpendicular distance from the roll axis.
(VI) estimating the centroid side deflection angle and the tire side force of the vehicle based on an interactive multi-model algorithm-volume Kalman filtering, wherein the volume Kalman filtering process comprises the following steps:
discretizing a state equation and an observation equation of a mass center slip angle and a tire lateral force of the distributed driving electric automobile:
wherein x is k ∈R n Is a state vector of the system, R n Is an n-dimensional real number set; u. of k ∈R m For known control inputs, R m Is a m-dimensional real number set; z is a radical of formula k ∈R p Is the observation vector of the system, R p Is a p-dimensional real number set; function f R n ×R m →R n And h is R n ×R m →R p Respectively known nonlinear functions, and the arrow is a nonlinear mapping relation; w is a k And v k Respectively, the process noise and the observed measurement noise of the system;
initialization:
wherein,
and P
0 Is the initial system state and error covariance; />
And &>
The mean and variance of the initial noise of the system process; />
And &>
Initially measuring a noise mean and variance for a system process;
updating time:
will be provided with
As an initial state of the covariance of model j, <' >>
Initial state as a mixture state estimate for model j, both as P for volumetric Kalman filtering
k-1|k-1 And &>
Initial input, factorization of error covariance matrix P
k-1|k-1 :
P k-1|k-1 =S k-1|k-1 S T k-1|k-1
Wherein S is k-1|k-1 Is a lower triangular matrix;
calculating volume points
Wherein,
k =2n, n being the dimension of the state to be estimated;
State estimation
Expression:
error covariance matrix P xx,k|k-1 Expressed as:
wherein Q is the system process noise variance;
updating measurement:
error covariance matrix P after factorization update k | k-1 :
Wherein S is k|k-1 Is a lower triangular matrix;
Volume point of propagation
Observation of predicted values
Estimated as:
innovation variance matrix P zz,k|k-1 The expression is as follows:
estimating a covariance matrix P xz,k|k-1 The expression is as follows:
kalman gain W k Is expressed as
Status update
The expression is as follows:
error covariance P xx,k|k The expression is as follows:
(VII) estimating the mass center slip angle and the tire lateral force of the vehicle based on an interactive multi-model algorithm-volume Kalman filtering, wherein the process of updating the model probability is as follows:
updating the model probability:
and updating the model probability by adopting a likelihood function, wherein the likelihood function of the model j is as follows:
wherein v is
j (k) Expressed as an innovation matrix, S
j (k) The covariance matrix of the innovation is represented,
the inverse of the innovation covariance matrix, <' >>
As a transpose of the innovation matrix, Λ
j (k) Representing the likelihood function of model j. />
The probability update for model j is:
wherein c represents a normalization constant, and
Λ
j (k) A likelihood function representing model j, <' >>
Represents the prediction probability, μ, of model j
j (k) Probability of model j.
(VIII) estimating the centroid slip angle and the tire lateral force of the vehicle based on an interactive multi-model algorithm-volume Kalman filtering, wherein the output interactive process comprises the following steps:
and (4) outputting interaction:
based on the model probability, the estimation results of all the filters are weighted and summed, and finally, the state estimation is calculated
The following can be obtained:
wherein, mu
j (k) The probability of the model j is determined,
based on the result of the state estimation of the volume Kalman>
Is the final state estimation result;
based on the model probabilities, the estimation results of all filters are weighted and summed, and finally the covariance estimation P (k | k) is calculated:
wherein, mu
j (k) The probability of the model j is determined,
as a result of the state estimation of the volumetric Kalman, P
j (k | k) is the volume Kalman like covariance estimate, P (k | k) is the final covariance estimate, and->
Is the final state estimation result.
Compared with the prior art, the invention has the following technical advantages:
the method estimates the vehicle centroid side deviation angle and the tire side force in real time based on the interactive multi-model algorithm-volume Kalman filtering, aims at the uncertainty of nonlinear and non-established dynamic models, and leads to vehicle state parameter estimation errors due to inaccurate modeling, can perform weighted calculation on the estimation results of two different vehicle road system models, and fully utilizes the estimation results of the different vehicle road system models under different working conditions, so that the problem of inaccurate estimation results due to inaccurate non-linear dynamic model establishment is greatly reduced, the Kalman filtering reaches three-order approximation, and the estimation accuracy of the vehicle centroid side deviation angle and the tire side force is improved.
Detailed Description
The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic views illustrating only the basic structure of the present invention in a schematic manner, and thus show only the constitution related to the present invention.
The method for estimating the centroid slip angle and the tire lateral force of the distributed driving electric automobile comprises the following steps as shown in the figure:
step one, establishing an eight-degree-of-freedom whole vehicle dynamics model of a vehicle, as shown in figures 2-4;
considering longitudinal motion, transverse motion, yaw motion, roll motion and motion of four tires of a vehicle under a complex working condition, and establishing a vehicle dynamic model;
the eight-degree-of-freedom vehicle dynamics equation is established as follows:
wherein, the expressions (1) and (2) are respectively longitudinal and transverse motion equations of the vehicle body of the vehicle, the expressions (3) and (4) are respectively longitudinal and transverse motion equations of the whole vehicle, and the expressions (5) and (6) are respectively transverse and side-rolling motion equations of the whole vehicle; m, m s Respectively representing the total mass of the vehicle and the sprung mass of the vehicle; i is zz 、I xz 、I xxs 、I xzs Respectively representing the yaw moment of the vehicle, the product of the moment of inertia of the vehicle mass around the x and z axes, the yaw moment of the vehicle sprung mass and the product of the moment of inertia of the vehicle sprung mass around the x and z axes; v x 、V y 、ω z 、ω x Respectively representing a vehicle longitudinal speed, a vehicle lateral speed, a vehicle yaw rate and a vehicle yaw rate; beta, delta and phi respectively represent a vehicle mass center slip angle, a front wheel rotating angle and a vehicle slip angle; f x 、F y Respectively represent longitudinal force and lateral force of four wheels of the vehicle(ii) a Subscripts i = f, r respectively denote a vehicle front axle and a vehicle rear axle, subscripts j = q, p respectively denote a vehicle left wheel and a vehicle right wheel, whereby subscripts fq, fp, rq, rp respectively denote a vehicle left front wheel, a right front wheel, a left rear wheel, and a right rear wheel; h represents the vertical distance from the center of mass of the sprung part of the vehicle to the roll axis; m z 、M x Respectively showing yaw moment and roll moment; b f 、B r Respectively showing the width of the front and the rear vehicle rails of the vehicle; l is f 、L r Respectively representing the front and rear wheel base of the vehicle; g represents the acceleration of gravity; the upper symbol "·" represents the differentiation of the indicated quantity.
Step two, selecting a linear tire model and a nonlinear Dugoff tire model as a model set of an interactive multi-model algorithm
Establishing an expression of vertical load of each tire as follows:
wherein, F z Representing vehicle vertical forces; subscript i = f, r represents the vehicle front and rear axles, respectively, subscript j = q, p represents the vehicle left and right wheels, respectively, whereby subscripts fq, fp, rq, rp represent the vehicle left and right front wheels, respectively; m, m s 、m uf 、m ur Respectively representing the total mass of the vehicle, the sprung mass of the vehicle, and the front and rear unsprung masses of the vehicle; c φf 、C φr Respectively representing the front roll rigidity and the rear roll rigidity of the vehicle; k φf 、K φr Respectively representing the front roll damping coefficient and the rear roll damping coefficient of the vehicle; v x 、V y 、ω z 、ω x Respectively representing a vehicle longitudinal speed, a vehicle lateral speed, a vehicle yaw rate and a vehicle yaw rate; g represents the gravitational acceleration; h is cg 、h uf 、h ur Respectively representing the mass center of the vehicle and the heights of the front and rear wheel side eccentric centers from the ground; respectively representing the front and rear roll center heights of the vehicle; b is f 、B r Respectively showing the front and rear rail widths of the vehicle; l is f 、L r Respectively representing the front and rear wheel base of the vehicle; phi represents a roll angle; the upper symbol "·" denotes the differentiation of the quantity indicated.
The tire slip ratio expression is:
wherein λ represents a tire slip ratio; v x 、ω ij Respectively representing the vehicle longitudinal speed and the tire yaw rate; subscripts i = f, r respectively denote a vehicle front axle and a vehicle rear axle, subscripts j = q, p respectively denote a vehicle left wheel and a vehicle right wheel, whereby subscripts fq, fp, rq, rp respectively denote a vehicle left front wheel, a right front wheel, a left rear wheel, and a right rear wheel; r e Is the effective radius of the tire.
(II) selecting a linear tire model and a nonlinear Dugoff tire model as a model set of an interactive multi-model algorithm;
the linear tire model is as follows:
the linear tire model slip angle expression is:
wherein α and δ respectively represent a tire slip angle and a front wheel rotation angle; v x 、V y 、ω z Respectively representing a vehicle longitudinal speed, a vehicle lateral speed and a vehicle yaw rate; b is f 、B r Respectively showing the front and rear rail widths of the vehicle; l is f 、L r Respectively representing the front and rear wheel base of the vehicle; phi represents a roll angle; epsilon f 、ε r Respectively representing the side tilting rotation coefficients of the front shaft and the rear shaft; subscripts i = f, r respectively denote a vehicle front axle and a vehicle rear axle, subscripts j = q, p respectively denote a vehicle left wheel and a vehicle right wheel, whereby subscripts fq, fp, rq, rp respectively denote a vehicle left front wheel, a right front wheel, a left rear wheel, and a right rear wheel.
The linear tire longitudinal and lateral force expressions are:
F yij =-C yij α ij
F xij =k μ λ ij F zij
wherein, F x 、F y Respectively representing the longitudinal force and the transverse force of four wheels of the vehicle; alpha and lambda respectively represent a tire slip angle and a slip ratio; c y 、k μ Indicating the cornering stiffness and the linear region mu-lambda of the tyre ij An image slope; subscripts i = f, r respectively denote a vehicle front axle and a vehicle rear axle, subscripts j = q, p respectively denote a vehicle left wheel and a vehicle right wheel, whereby subscripts fq, fp, rq, rp respectively denote a vehicle left front wheel, a right front wheel, a left rear wheel, and a right rear wheel.
The non-linear Dugoff tire model is as follows:
the nonlinear tire model slip angle expression is as follows:
wherein, alpha and delta represent a tire slip angle and a front wheel rotation angle; v x 、V y 、ω z Respectively representing a vehicle longitudinal speed, a vehicle transverse speed and a vehicle yaw rate; b is f 、B r Respectively showing the width of the front and the rear vehicle rails of the vehicle; l is f 、L r Respectively representing the front and rear wheel base of the vehicle; phi represents a roll angle; subscripts i = f, r respectively denote a vehicle front axle and a vehicle rear axle, subscripts j = q, p respectively denote a vehicle left wheel and a vehicle right wheel, whereby subscripts fq, fp, rq, rp respectively denote a vehicle left front wheel, a right front wheel, a left rear wheel, and a right rear wheel.
The non-linear Dugoff tire longitudinal and lateral force expressions are:
wherein, F x 、F y Respectively representing the longitudinal force and the transverse force of four wheels of the vehicle; alpha and lambda represent the slip angle and slip ratio of the tire; c y Representing the tire cornering stiffness; s is the substitution amount of the function; subscript i = f, r represents the vehicle front and rear axles, respectively, subscript j = q, p represents the vehicle left and right wheels, respectively, whereby subscripts fq, fp, rq, rp represent the vehicle left and right front, rear, and rear wheels, respectively.
Estimating the mass center and the side deflection angle of the vehicle and the tire side force based on an interactive multi-model algorithm-volume Kalman filtering, wherein the algorithm process comprises the following steps: input interaction, volumetric kalman filtering, update model probability, and output interaction, as shown in fig. 5;
the input interaction process is as follows:
the model probability mu of the model j in the previous step
j (k-1) and State estimation
Gets a mixed estimate->
Assuming that the linear tire model and the Dugoff nonlinear tire model are models i, j =1,2, respectively, the transfer matrix is therefore:
wherein P is a transition matrix, P in the transition matrix ij The element is the motion transition probability of the target from the ith model to the jth model, the subscript r = j of the element in the transition matrix P, and r is the number of model set models.
The prediction model is derived from the expression of model j:
wherein, mu
j (k-1) represents the model probability of model j at time k-1 of the previous step,
the prediction probability of the model j is represented,
the summation operation for model i from 1 to r.
The expression of the mixing probability from model i to model j is:
wherein, mu ij (k-1; k-1) represents the mixing probability from model i to model j.
The hybrid state estimate is derived from the expression of model j:
wherein,
is a state estimate of the target, based on the status of the target>
Is a hybrid state estimate of model j.
Model probability mu of all filters in the last step
j (k-1) and State estimation
Get the covariance->
The hybrid covariance estimate is derived from model j as:
where T represents the transpose of the matrix,
representing the initial state of the covariance of model j, device for combining or screening>
Representing the covariance of model i.
(II) establishing distributed driving electric automobile model
Establishing a state equation and an observation equation of a mass center slip angle and a tire lateral force of the distributed driving electric automobile:
wherein,
is the first derivative of the state variable; f (-) is the vehicle model equation of state; h (-) is the vehicle model observation equation; x (t) is a state variable; u (t) is an input variable; z (t) isMeasuring a variable; w (t) and v (t) are zero-mean, uncorrelated white noise;
x(t)=(β,F yfq ,F yfp ,F yrq ,F yrp ) T
u(t)=(δ,F xfq ,F xfp ,F xrq ,F xrp ,ω fq ,ω fp ,ω rq ,ω rp ) T
z(t)=(a x ,a y ,ω z ,ω x ) T
wherein, F x 、F y Respectively representing the longitudinal force and the transverse force of four wheels of the vehicle; alpha, delta and beta respectively represent a tire slip angle, a front wheel corner and a centroid slip angle; omega z 、ω x 、ω ij Respectively representing the yaw angular velocity, the vehicle roll angle velocity and the wheel rotating speed of the vehicle; a is x 、a y Respectively representing the longitudinal acceleration and the lateral acceleration of the vehicle; subscripts i = f, r respectively denote a vehicle front axle and a vehicle rear axle, subscripts j = q, p respectively denote a vehicle left wheel and a vehicle right wheel, whereby subscripts fq, fp, rq, rp respectively denote a vehicle left front wheel, a right front wheel, a left rear wheel, and a right rear wheel.
Establishing a vehicle model state equation f (-) and an observation equation h (-) by:
the functional expression f (-) is:
the functional expression h (-) is:
wherein, f 1 -f 5 Respectively representing the state equations of the vehicle model; h is 1 -h 4 Respectively representing vehicle model observation equations; f x 、F y Respectively representing the longitudinal force and the transverse force of four wheels of the vehicle; alpha, delta and beta respectively represent a tire slip angle, a front wheel corner and a centroid slip angle; λ represents a tire slip ratio; lower partThe index i = f, r denotes the vehicle front axle and the vehicle rear axle, respectively, the index j = q, p denotes the vehicle left wheel and the vehicle right wheel, respectively, whereby the indices fq, fp, rq, rp denote the vehicle left front wheel, right front wheel, left rear wheel and right rear wheel, respectively; m, m s Respectively representing the total mass of the vehicle and the sprung mass of the vehicle; omega z 、ω x Respectively representing the vehicle yaw angular velocity and the vehicle side slip angular velocity; g represents the gravitational acceleration; phi represents a roll angle; h represents the vertical distance from the center of mass of the sprung portion of the vehicle to the roll axis.
(III) the process of the volume Kalman filtering is as follows:
discretizing a state equation and an observation equation of a mass center slip angle and a tire lateral force of the distributed driving electric automobile:
wherein x is k ∈R n Is a state vector of the system, R n Is an n-dimensional real number set; u. u k ∈R m For known control inputs, R m Is a m-dimensional real number set; z is a radical of k ∈R p Is the observation vector of the system, R p Is a p-dimensional real number set; function f R n ×R m →R n And h is R n ×R m →R p Respectively known nonlinear functions, and the arrow is a nonlinear mapping relation; w is a k And v k Respectively, the process noise and the observed measurement noise of the system;
initialization:
wherein,
and P
0 Is the initial system state and error covariance; />
And &>
The mean and variance of the initial noise of the system process; />
And &>
Initially measuring a noise mean and variance for a system process;
updating time:
will be provided with
As an initial state of the covariance of model j, <' >>
Initial state as a mixture state estimate for model j, both as P for volumetric Kalman filtering
k-1|k-1 And &>
Initial input, factorized error covariance matrix P
k-1|k-1 :
P k-1|k-1 =S k-1|k-1 S T k-1|k-1
Wherein S is k-1k-1 Is a lower triangular matrix;
calculating volume points
Wherein,
k =2n, n being the dimension of the state to be estimated;
State estimation
Expression: />
Error covariance matrix P xx,k|k-1 Expressed as:
wherein Q is the system process noise variance;
updating measurement:
error covariance matrix P after factorization update k|k-1 :
Wherein S is k|k-1 Is a lower triangular matrix;
Volume point of propagation
Observation of predicted values
Estimated as:
innovation variance matrix P zz,k|k-1 The expression is as follows:
estimating a covariance matrix P xz,kk-1 The expression is as follows:
kalman gain W k Is expressed as
Status update
The expression is as follows:
error covariance P xx,k|k The expression is as follows:
(IV) the process of updating the model probabilities is as follows:
updating the model probability:
and updating the model probability by adopting a likelihood function, wherein the likelihood function of the model j is as follows:
wherein v is
j (k) Expressed as an innovation matrix, S
j (k) The covariance matrix of the innovation is represented,
the inverse of the innovation covariance matrix, <' >>
As a transpose of the innovation matrix, Λ
j (k) Representing the likelihood function of model j.
The probability update for model j is:
wherein c represents a normalization constant, and
Λ
j (k) A likelihood function representing model j, <' >>
Represents the prediction probability, μ, of model j
j (k) Probability of model j.
(V) the process of output interaction is as follows:
and (4) outputting interaction:
based on the model probability, the estimation results of all the filters are weighted and summed, and finally, the state estimation is calculated
The following can be obtained:
wherein, mu
j (k) The probability of the model j is determined,
based on the result of the state estimation of the volume Kalman>
Is the final state estimation result;
based on the model probabilities, the estimation results of all filters are weighted and summed, and finally the covariance estimation P (k | k) is calculated:
wherein, mu
j (k) The probability of the model j is determined,
as a result of the state estimation of the volumetric Kalman, P
j (k | k) is the volume Kalman like covariance estimate, and P (k | k) is the final covariance estimate, based on>
Is the final state estimation result. />