CN113253615B

CN113253615B - Motion state observation method and system based on distributed electric chassis

Info

Publication number: CN113253615B
Application number: CN202110694684.XA
Authority: CN
Inventors: 张冰; 王心醉; 梁岳青; 李鹏飞; 李希华; 李兆波; 蔡后勇; 辛海兵
Original assignee: Individual
Current assignee: Individual
Priority date: 2021-06-22
Filing date: 2021-06-22
Publication date: 2021-10-22
Anticipated expiration: 2041-06-22
Also published as: CN113253615A

Abstract

The invention relates to the technical field of information processing, and discloses a motion state observation method and a motion state observation system based on a distributed electric chassis. In order to avoid the influence of positive qualitative damage caused by introducing multiple fading factors on the observation precision of the motion state, positive parameters are respectively introduced into the two observation layers to improve the two observation layers, so that the accuracy of the motion state observation under the drive fault is further ensured.

Description

Motion state observation method and system based on distributed electric chassis

Technical Field

The invention relates to the technical field of information processing, in particular to a motion state observation method and system based on a distributed electric chassis.

Background

The distributed electric chassis has four-wheel independent driving, braking and steering functions, can realize more flexible and comfortable dynamics and kinematics control through a reasonable control means (wherein, the distributed electric chassis based on hub motor driving is widely applied to various industries, such as distributed electric automobiles, distributed electric wheelchairs and the like), and provides an ideal carrier for intellectualization and convenience of related products. Accurate acquisition of chassis motion state information is an important technical basis for realizing intellectualization and dynamics control of the chassis motion state information, however, an accurate motion state signal is difficult to obtain directly through a hardware sensor, and a motion state observer needs to be established. The distributed electric chassis motion state observer designed by the traditional method has poor robustness to model uncertainty, has poor observation capability to slow change or mutation of the complete machine motion state caused by the failure of a hub motor or other reasons after entering a steady state, and sets the same observation performance for different distributed electric chassis complete machine motion states, thereby limiting the observation capability to different motion states. Therefore, how to solve the problem that the existing motion state observer based on the distributed electric chassis cannot quickly and accurately observe different motion states when a drive fault occurs becomes an urgent solution.

The above is only for the purpose of assisting understanding of the technical aspects of the present invention, and does not represent an admission that the above is prior art.

Disclosure of Invention

The invention mainly aims to provide a motion state observation method and system based on a distributed electric chassis, and aims to solve the technical problem that the existing motion state observer based on the distributed electric chassis cannot quickly and accurately observe different motion states when a drive fault occurs.

In order to achieve the above object, the present invention provides a motion state observation method based on a distributed electric chassis, comprising the following steps:

respectively constructing a complete machine observation model corresponding to a complete machine motion state observation layer and a driving observation model corresponding to a driving state observation layer in a motion state observer based on a distributed electric chassis;

calculating a complete machine prior estimation value corresponding to the complete machine motion state observation layer at the current moment according to the complete machine observation model, and calculating a driving prior estimation value corresponding to the driving state observation layer at the current moment according to the driving observation model;

respectively calculating a complete machine multiple time-varying fading factor matrix corresponding to the complete machine motion state observation layer and a driving multiple time-varying fading factor matrix corresponding to the driving state observation layer;

calculating a complete machine prior forecast variance matrix corresponding to the complete machine motion state observation layer according to the complete machine multiple time-varying fading factor matrix and the improved complete machine covariance matrix, and calculating a drive prior forecast variance matrix corresponding to the drive motion state observation layer according to the drive multiple time-varying fading factor matrix and the improved drive covariance matrix;

calculating a complete machine state correction gain corresponding to the complete machine motion state observation layer according to the complete machine priori prediction variance matrix, and calculating a driving state correction gain corresponding to the driving state observation layer according to the driving priori prediction variance matrix;

correcting the complete machine prior estimated value according to the complete machine state correction gain to obtain a complete machine optimal estimated value, and correcting the drive prior estimated value according to the drive state correction gain to obtain a drive optimal estimated value;

and observing the motion state at the next moment based on the complete machine optimal estimation value, the drive optimal estimation value, the complete machine prior prediction variance matrix and the drive prior prediction variance matrix.

Optionally, the step of respectively constructing a complete machine observation model corresponding to a complete machine motion state observation layer and a drive observation model corresponding to a drive state observation layer in the motion state observer based on the distributed electric chassis specifically includes:

respectively establishing a complete machine dynamic model, a tire cornering model and a wheel dynamic model according to the structural characteristics of the distributed electric chassis;

acquiring a preset observation vector according to the complete machine dynamic model, the tire lateral deviation model and the wheel dynamic model, and generating a complete machine observation model corresponding to a complete machine motion state observation layer according to the preset observation vector through the following formula, wherein the preset observation vector comprises a complete machine motion state vector, a driving state variable, a joint state input vector and a joint state measurement vector,

in the formula, x_vic(k + 1) is the complete machine motion state vector of the complete machine motion state observation layer at the current moment, x_vic(k) The complete machine motion state vector of the complete machine motion state observation layer at the previous moment, F_vic(k) A corresponding system state Jacobian matrix H after linearization of a complete machine observation model at the previous moment_vic(k + 1) is the measured Jacobian matrix u corresponding to the linearized whole observation model at the current moment_DMFSTF(k) As input vector of the observer of the state of motion at the previous moment, U_vicIs an input vector, x, of an observation layer of the motion state of the whole machine_div(k) Observing the driving state vector of the layer for the driving state of the previous moment, Q_vic(k) Covariance, y, of process noise at the previous moment observation layer for the overall machine motion state_DMFSTF(k + 1) is the measurement vector of the observer of the state of motion at the current moment, R_vic(k + 1) is the covariance of the measurement noise of the observation layer of the complete machine motion state at the current moment;

converting the complete machine observation model according to the relation between the complete machine motion state observation layer and the driving state observation layer by the following formula to obtain a driving observation model corresponding to the driving state observation layer,

in the formula, x_div(k + 1) is a drive state vector of the drive state observation layer at the current time, x_div(k) Observing the driving state vector of the layer for the driving state of the previous moment, Q_div(k) Covariance of process noise, y, for the observation layer of the drive state at the previous time_DMFSTF(k + 1) is the measurement vector of the observer of the state of motion at the current moment, H_divAnd (k + 1) is a measured Jacobian matrix of the driving observation model at the current moment.

Optionally, the step of calculating a complete machine prior estimated value corresponding to the complete machine motion state observation layer at the current time according to the complete machine observation model, and calculating a drive prior estimated value corresponding to the drive state observation layer at the current time according to the drive observation model specifically includes:

calculating a complete machine prior estimated value corresponding to the complete machine motion state observation layer at the current moment according to the complete machine observation model by the following formula,

in the formula (I), the compound is shown in the specification,

(k +1 | k) is a global priori estimated value corresponding to the global motion state observation layer at the current time predicted from the global priori estimated value corresponding to the global motion state observation layer at the previous time, F_vic(k) The Jacobian matrix of the system state, x, after the linearization of the whole machine observation model at the previous moment_vic(k) The complete machine motion state vector, U, of the complete machine motion state observation layer at the previous moment_vicIs an input vector u of the observation layer of the motion state of the whole machine_DMFSTF(k) Is the input quantity, x, of the observer of the state of motion at the previous moment_div(k) Observing the driving state vector of the layer for the driving state of the previous moment, Q_vic(k) The covariance of the process noise of the whole machine motion state observation layer at the previous moment is obtained;

calculating a driving prior estimation value corresponding to the driving state observation layer at the current moment according to the driving observation model by the following formula,

in the formula (I), the compound is shown in the specification,

(k +1 | k) is a drive prior estimated value corresponding to the drive state observation layer at the current time predicted from the drive prior estimated value corresponding to the drive state observation layer at the previous time,

(k) the driving prior estimation value is corresponding to the driving state observation layer at the previous moment.

Optionally, the step of calculating a complete machine multiple time-varying fading factor matrix corresponding to the complete machine motion state observation layer and a driving multiple time-varying fading factor matrix corresponding to the driving state observation layer respectively specifically includes:

calculating a complete machine multiple time-varying fading factor matrix corresponding to the complete machine motion state observation layer through the following formula,

in the formula, λ_vicIs a complete machine multiple time-varying fading factor matrix, lambda, corresponding to the observation layer of the complete machine motion state_{vic_1}、λ_{vic_2}……λ_{vic_n}Multiple time-varying fading factors x for different overall motion states in the overall motion state observation layer_vic（1）、x_vic（1）……x_vic(n) different overall motion state vectors, lambda, of the overall motion state observation layer_{vic_i}(k + 1) is a multiple time-varying fading factor, eta, of the whole machine corresponding to the observation layer of the whole machine motion state at the current moment_{vic_i}(k + 1) is a preset multiple parameter of the complete machine motion state observation layer at the current moment;

calculating a driving multiple time-varying fading factor matrix corresponding to the driving state observation layer by the following formula,

in the formula, λ_divFor driving multiple time-varying fading factor matrix, lambda, corresponding to the driving state observation layer_{div_T_lf}Multiple time-varying fading factors, lambda, corresponding to the driving torque of the hub motor of the left front wheel_{div_T_rf}Is a multiple time-varying fading factor corresponding to the driving torque of the hub motor of the right front wheel,λ_{div_T_lr}multiple time-varying fading factors, lambda, corresponding to the driving torque of the hub motor of the left rear wheel_{div_T_rr}Multiple time-varying fading factors corresponding to the driving torque of the hub motor of the right rear wheel, T_{_lf}Driving torque, T, of a hub motor for the left front wheel_{_rf}Driving torque, T, of a hub motor for the right front wheel_{_lr}Driving torque, T, of a hub motor for a left rear wheel_{_rr}Driving torque, lambda, of a hub motor for the right rear wheel_{div_i}(k + 1) is a driving multiple time-varying fading factor, η, corresponding to the driving state observation layer at the current time_{div_i}And (k + 1) is a preset multiple parameter of the driving state observation layer at the current moment.

Optionally, the preset multi-parameter matrix of the overall machine motion state observation layer at the current moment is obtained by the following formula,

in the formula eta_vicFor presetting multiple parameter matrixes, eta, of the observation layer of the motion state of the whole machine_{vic_1}、η_{vic_2}……η_{vic_n}Different multi-parameter matrices, eta, for the observation layer of the state of motion of the machine_vic(k + 1) is a preset multi-parameter matrix of the driving state observation layer at the current moment, F_vic(k) A system state Jacobian matrix H corresponding to the linearized whole machine observation model at the previous moment_vic(k + 1) is a measured Jacobian matrix, Q, corresponding to the linearized whole machine observation model at the current moment_vic(k) Covariance of process noise, beta, of the observation layer of the complete machine motion state at the previous moment_vicMultiple weakening factors, R, for the observation layer of the state of motion of the whole machine_vic(k + 1) is the covariance of the measurement noise of the observation layer of the complete machine motion state at the current moment, P_vic(k) Forecasting the variance matrix, gamma, for the whole machine prior corresponding to the observation layer of the whole machine motion state at the previous moment_vicIs the residual error of the observation layer of the motion state of the whole machine, V_{0_vic}(k + 1) is a residual covariance matrix of the observation layer of the complete machine motion state at the current moment, V_{0_vic}(k) Is a residual covariance matrix, rho, of the observation layer of the complete machine motion state at the previous moment_vicMultiple forgetting factor, gamma, of the observation layer of the motion state of the whole machine_vic(k + 1) is the residual error of the observation layer of the complete machine motion state at the current moment;

accordingly, the preset multi-parameter matrix of the driving state observation layer at the current moment is obtained by the following formula,

in the formula eta_divPredetermined multiple parameter matrix, η, for driving the state observation layers_{div_T_lf}Is a multi-parameter matrix, eta, corresponding to the driving torque of the hub motor of the left front wheel_{div_T_rf}A multi-parameter matrix, eta, corresponding to the driving torque of the hub motor of the right front wheel_{div_T_lr}A multi-parameter matrix, eta, corresponding to the driving torque of the hub motor of the left rear wheel_{div_T_rr}A matrix of multiple parameters, eta, corresponding to the driving torque of the hub motor of the right rear wheel_div(k + 1) is a preset multi-parameter matrix of the driving state observation layer at the current moment, F_div(k) The Jacobian matrix of the system states, H, corresponding to the driving observation model at the previous moment_div(k + 1) is a measured Jacobian matrix corresponding to the driving observation model at the current moment, V_{0_div}(k) Residual covariance matrix, V, for the observation layer of the drive state at the previous moment_{0_div}(k + 1) is the residual covariance matrix of the observation layer of the drive state at the previous time, Q_div(k) Covariance of process noise, beta, for the observation layer of the driving state at the previous moment_divMultiple attenuation factors, P, for driving the state observation layers_div(k) Predicting a variance matrix, gamma, for a drive prior corresponding to the observation layer of the drive state at the previous moment_divFor the residual of the observation layer in the driven state, p_divFor observing layers for driving statesMultiple forgetting factor, gamma_div(k + 1) is the residual of the drive state observation layer at the current time.

Optionally, the step of calculating a complete machine priori prediction variance matrix corresponding to the complete machine motion state observation layer according to the complete machine multiple time-varying fading factor matrix and the improved complete machine covariance matrix, and calculating a drive priori prediction variance matrix corresponding to the drive motion state observation layer according to the drive multiple time-varying fading factor matrix and the improved drive covariance matrix specifically includes:

calculating a complete machine prior forecast variance matrix corresponding to the complete machine motion state observation layer according to the complete machine multiple time-varying fading factor matrix and the improved complete machine covariance matrix through the following formula,

in the formula, P_vic(k +1 | k) is a total machine prior prediction variance matrix corresponding to the total machine motion state observation layer at the current time predicted according to the total machine prior prediction variance matrix corresponding to the total machine motion state observation layer at the previous time, and P is_vic(k) Forecasting the variance matrix, lambda, of the whole machine prior corresponding to the observation layer of the whole machine motion state at the previous moment_vic(k + 1) is a complete machine multiple time-varying fading factor matrix corresponding to the complete machine motion state observation layer at the current moment, F_vic(k) A system state Jacobian matrix, Q, corresponding to the linearized whole machine observation model at the previous moment_vic(k) The covariance of the process noise of the whole machine motion state observation layer at the previous moment is obtained;

calculating a driving prior forecast variance matrix corresponding to the driving motion state observation layer according to the driving multiple time-varying fading factor matrix and the improved driving covariance matrix by the following formula,

in the formula, P_div(k +1 | k) is a drive prior prediction variance matrix corresponding to the drive state observation layer at the current time predicted from the drive prior prediction variance matrix corresponding to the drive state observation layer at the previous time, P_div(k) Predicting a variance matrix, lambda, for a drive prior corresponding to the observation layer of the drive state at the previous time_div(k + 1) is a driving multiple time-varying fading factor matrix, Q, corresponding to the driving state observation layer at the current time_div(k) The covariance of the process noise of the layer is observed for the drive state at the previous time instant.

Optionally, the step of calculating a complete machine state correction gain corresponding to the complete machine motion state observation layer according to the complete machine prior prediction variance matrix, and calculating a driving state correction gain corresponding to the driving state observation layer according to the driving prior prediction variance matrix specifically includes:

calculating the whole machine state correction gain corresponding to the whole machine motion state observation layer according to the whole machine prior prediction variance matrix through the following formula,

in the formula, K_vic(k + 1) is the overall machine state correction gain, P, corresponding to the overall machine motion state observation layer_vic(k +1 | k) is a total machine prior prediction variance matrix corresponding to the total machine motion state observation layer at the current time predicted according to the total machine prior prediction variance matrix corresponding to the total machine motion state observation layer at the previous time, H_vic(k + 1) is a measured Jacobian matrix corresponding to the linearized whole machine observation model at the current moment, R_vic(k) The covariance of the measurement noise of the observation layer of the complete machine motion state at the previous moment;

calculating a driving state correction gain corresponding to the driving state observation layer according to the driving prior forecast variance matrix by the following formula,

in the formula, K_div(k + 1) is a drive state correction gain corresponding to the drive state observation layer, P_div(k +1 | k) is a drive prior prediction variance matrix corresponding to the drive state observation layer at the current time predicted from the drive prior prediction variance matrix corresponding to the drive state observation layer at the previous time, H_divAnd (k + 1) is a measured Jacobian matrix of the driving observation model at the current moment.

Optionally, the step of correcting the complete machine prior estimated value according to the complete machine state correction gain to obtain a complete machine optimal estimated value, and correcting the drive prior estimated value according to the drive state correction gain to obtain a drive optimal estimated value specifically includes:

correcting the complete machine prior estimated value according to the complete machine state correction gain by the following formula to obtain the complete machine optimal estimated value,

in the formula (I), the compound is shown in the specification,

(k + 1) is the optimal estimation value of the whole machine corresponding to the observation layer of the whole machine motion state at the current moment,

(K +1 | K) is a global priori estimated value corresponding to the global motion state observation layer at the current time predicted according to the global priori estimated value corresponding to the global motion state observation layer at the previous time, K_vic(k + 1) is the overall machine state correction gain corresponding to the overall machine motion state observation layer, y_DMFSTF(k + 1) is the measurement vector of the observer of the state of motion at the current moment, H_vic(k + 1) is a system state Jacobian matrix corresponding to the linearized complete machine observation model at the current moment;

correcting the driving prior estimated value according to the driving state correction gain by the following formula to obtain a driving optimal estimated value,

in the formula (I), the compound is shown in the specification,

(k + 1) is the driving optimal estimation value corresponding to the driving state observation layer at the current moment,

(K +1 | K) is a drive prior estimate value corresponding to the current-time drive state observation layer predicted from the drive prior estimate value corresponding to the previous-time drive state observation layer, and K_div(k + 1) is a drive state correction gain corresponding to the drive state observation layer, H_divAnd (k + 1) is a measured Jacobian matrix of the driving observation model at the current moment.

Optionally, after obtaining the complete machine prior prediction variance matrix corresponding to the complete machine motion state observation layer, the method further includes:

correcting the whole machine prior forecast variance matrix corresponding to the whole machine motion state observation layer through the following formula to obtain a whole machine optimal variance matrix corresponding to the whole machine motion state observation layer,

in the formula, P_vic(k + 1) is the optimal variance matrix of the whole machine corresponding to the observation layer of the whole machine motion state at the current moment, F_vic(k) A system state Jacobian matrix, P, corresponding to the linearized whole machine observation model at the previous moment_vic(k +1 | k) is a total machine prior prediction variance matrix corresponding to the total machine motion state observation layer at the current time predicted according to the total machine prior prediction variance matrix corresponding to the total machine motion state observation layer at the previous time, and Q is_vic(k) For the complete machine operation at the previous momentCovariance of process noise at the dynamic state observation layer,

_{vic_1}、

_{vic_2}、……

_{vic_n}for different positive parameters, K, of the observation layer of the state of motion of the machine_vic(k + 1) is the overall machine state correction gain corresponding to the overall machine motion state observation layer, H_vic(k + 1) is a measured Jacobian matrix, lambda, corresponding to the linearized overall observation model at the current moment_vic(k + 1) is a complete machine multiple time-varying fading factor matrix corresponding to the complete machine motion state observation layer at the current moment;

correspondingly, correcting the driving prior forecast variance matrix corresponding to the driving motion state observation layer by the following formula to obtain a driving optimal variance matrix corresponding to the driving state observation layer,

in the formula, P_div(k + 1) is the optimal variance matrix of the drive corresponding to the observation layer of the drive state at the current moment, F_div(k) The Jacobian matrix of the system state, P, corresponding to the driving observation model at the previous moment_div(k +1 | k) is a drive prior prediction variance matrix corresponding to the drive state observation layer at the current time predicted from the drive prior prediction variance matrix corresponding to the drive state observation layer at the previous time, Q_div(k) The covariance of the process noise of the layer is observed for the drive state at the previous time instant,

_{div_lf}positive definite parameters corresponding to the left front wheel in the driving state observation layer,

_{div_rf}Positive definite parameters corresponding to the right front wheel in the driving state observation layer,

_{div_lr}Positive definite parameters corresponding to the left rear wheel in the driving state observation layer,

_{div_rr}Positive definite parameter, K, corresponding to the right rear wheel in the observation layer of driving state_div(k + 1) is a drive state correction gain corresponding to the drive state observation layer, H_div(k + 1) is a measured Jacobian matrix corresponding to the driving observation model at the current moment, lambda_divAnd (k + 1) is a driving multiple time-varying fading factor matrix corresponding to the driving state observation layer at the current moment.

In addition, in order to achieve the above object, the present invention further provides a motion state observation system based on a distributed electric chassis, including:

the model construction module is used for respectively constructing a complete machine observation model corresponding to a complete machine motion state observation layer and a driving observation model corresponding to a driving state observation layer of the motion state observer based on the distributed electric chassis;

the prior value calculation module is used for calculating a complete machine prior estimated value corresponding to the complete machine motion state observation layer at the current moment according to the complete machine observation model and calculating a driving prior estimated value corresponding to the driving state observation layer at the current moment according to the driving observation model;

the factor calculation module is used for respectively calculating a complete machine multiple time-varying fading factor matrix corresponding to the complete machine motion state observation layer and a driving multiple time-varying fading factor matrix corresponding to the driving state observation layer;

the variance calculation module is used for calculating a complete machine priori forecast variance matrix corresponding to the complete machine motion state observation layer according to the complete machine multiple time-varying fading factor matrix and the improved complete machine covariance matrix, and calculating a drive priori forecast variance matrix corresponding to the drive motion state observation layer according to the drive multiple time-varying fading factor matrix and the improved drive covariance matrix;

the correction module is used for calculating the whole machine state correction gain corresponding to the whole machine motion state observation layer according to the whole machine priori prediction variance matrix and calculating the driving state correction gain corresponding to the driving state observation layer according to the driving priori prediction variance matrix;

the optimal value calculation module is used for correcting the complete machine prior estimated value according to the complete machine state correction gain to obtain a complete machine optimal estimated value, and correcting the drive prior estimated value according to the drive state correction gain to obtain a drive optimal estimated value;

and the recursive observation module is used for observing the motion state at the next moment based on the complete machine optimal estimation value, the drive optimal estimation value, the complete machine prior forecast variance matrix and the drive prior forecast variance matrix.

According to the invention, aiming at the motion state observation of the distributed electric chassis under the drive fault, a double-structure multiple fading factor strong tracking motion state observer is designed, a drive state observation layer and a complete machine motion state observation layer are respectively designed on the basis of a multiple fading factor strong tracking filter, different time-varying fading factors are respectively introduced aiming at each drive state and complete machine motion state, so that the different motion state observation requirements are met, and the fast and accurate tracking of the motion states of different distributed electric chassis under the drive fault is realized by combining the strong tracking of the states of the two observation layers and the data interaction between the two observation layers. Furthermore, in order to avoid the influence of positive qualitative damage caused by introducing multiple fading factors on the state observation precision of the whole distributed electric chassis, positive parameters are respectively introduced into the driving state observation layer and the whole machine motion state observation layer to improve the driving state observation layer and the whole machine motion state observation layer, and the accuracy of motion state observation under the driving fault is further ensured.

Drawings

FIG. 1 is a schematic flow chart of a first embodiment of a distributed electric chassis-based motion state observation method according to the present invention;

fig. 2 is a block diagram of a first embodiment of a motion state observation system based on a distributed electric chassis according to the present invention.

The implementation, functional features and advantages of the objects of the present invention will be further explained with reference to the accompanying drawings.

Detailed Description

It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

The embodiment of the invention provides a motion state observation method based on a distributed electric chassis, and referring to fig. 1, fig. 1 is a schematic flow diagram of a first embodiment of the motion state observation method based on the distributed electric chassis.

In this embodiment, the motion state observation method based on the distributed electric chassis includes the following steps:

step S10: respectively constructing a complete machine observation model corresponding to a complete machine motion state observation layer and a driving observation model corresponding to a driving state observation layer in a motion state observer based on a distributed electric chassis;

before a complete machine observation model corresponding to a complete machine (vehicle) motion state observation layer and a drive observation model corresponding to a drive (drive) state observation layer in the motion state observer based on the distributed electric chassis are established, a complete machine dynamic model, a tire cornering model and a wheel dynamic model can be respectively established according to the structural characteristics of the distributed electric chassis. In order to reduce the complexity of the model and improve the observation speed of the motion state, the complete machine dynamic model can be a three-degree-of-freedom dynamic model which is established based on Newton's second law and neglects pitching, side-tipping and vertical complete machine motion; the tire cornering power control method comprises the following steps that a tire cornering model is a model which is established by neglecting the influence of wheel heeling on tire force, based on a tire model of the Highway Safety Research Institute (HSRI), and is used for representing the relationship between a cornering angle and the cornering power of a tire, when the overall speed of a distributed electric chassis exceeds a certain threshold value, due to the material characteristics of the tire, the wheel is no longer a rigid object, due to the fact that the shape of the tire is distorted, the grounding part of the wheel cannot rotate along with other parts of the tire immediately, the distortion can enable the driving direction of the wheel and the plane axis of the tire to form an included angle, namely the cornering angle, and when the cornering angle is generated, the ground generates a lateral reaction force, namely the cornering power, to the tire; in the embodiment, the four wheels of the distributed electric chassis all adopt the hub motors as driving systems, and because the hub motors respond quickly and neglect the transient response process, the hub motors and the corresponding controllers can be regarded as an integral and independent sub-driving system, the output torque and the control input signal of the hub motors are in a direct proportion relation, and a model established based on a torque balance principle and the relation between the output torque and the control input signal is a wheel power model. The specific modeling modes of the complete machine dynamic model, the tire cornering model and the wheel dynamic model can be set according to actual requirements, so that the complete machine motion state vector, the driving state variable, the joint state input vector and the joint state measurement vector required by the embodiment can be quickly and accurately obtained, and the embodiment is not limited to this.

The complete machine motion state vector comprises but is not limited to a vector consisting of complete machine longitudinal acceleration, complete machine lateral acceleration, complete machine yaw angular velocity, complete machine yaw angular acceleration, complete machine longitudinal vehicle speed and lateral vehicle speed of the distributed electric chassis; the driving state variables include, but are not limited to, vectors composed of driving moments of four wheels (left front wheel, left rear wheel, right front wheel, right rear wheel) corresponding to the distributed electric chassis; the motion state observer related to the implementation can be understood as a Double-structure Multiple Fading factor Strong Tracking motion state observer generated based on a Double Multiple Fading factor Strong Tracking Filter (DMFSTF), and the basic idea is to respectively identify the motion state of the whole machine and different states of a driving system by adopting a two-layer Multiple Fading factor Strong Tracking filtering method, wherein different time-varying Fading factors are set for each driving state and the motion state of the whole machine while the driving fault of the distributed electric chassis is considered, so that various observation requirements of motion control on different states are met, and the problem of divergence of motion state information of the distributed electric chassis caused by the introduction of Multiple time-varying Fading factors is improved by introducing positive parameters; the self-adaptive accurate identification of different motion states of the distributed electric chassis is realized by combining strong tracking of abrupt change states of two observation layers and data interaction between the observation layers, and the method can be understood as a combined state observer, wherein state variables of a bottom actuator of the combined state observer are input of the complete machine observer, so the input of the actuator can be regarded as integral input of the combined state observer, and based on the integral input vector, the combined state input vector comprises but is not limited to four-wheel turning angles, wheel rotating speeds and rotating speed acceleration, and the combined state measurement vector comprises but is not limited to complete machine longitudinal acceleration, complete machine lateral acceleration and complete machine yaw angular speed of the distributed electric chassis.

In a specific implementation, a preset observation vector is obtained according to the complete machine dynamic model, the tire lateral deviation model and the wheel dynamic model, and a complete machine observation model corresponding to a complete machine motion state observation layer can be generated according to the preset observation vector through the following formula, wherein the preset observation vector comprises a complete machine motion state vector, a driving state variable, a joint state input vector and a joint state measurement vector,

in the formula, x_vic(k + 1) is the complete machine motion state vector of the complete machine motion state observation layer at the current moment, x_vic(k) The complete machine motion state vector of the complete machine motion state observation layer at the previous moment, F_vic(k) A corresponding system state Jacobian matrix H after linearization of a complete machine observation model at the previous moment_vic(k + 1) is the measured Jacobian matrix u corresponding to the linearized whole observation model at the current moment_DMFSTF(k) As input vector of the observer of the state of motion at the previous moment, U_vicFor observing the motion state of the whole machineInput vector of x_div(k) Observing the driving state vector of the layer for the driving state of the previous moment, Q_vic(k) Covariance, y, of process noise at the previous moment observation layer for the overall machine motion state_DMFSTF(k + 1) is the measurement vector of the observer of the state of motion at the current moment, R_vic(k + 1) is the covariance of the measurement noise of the observation layer of the complete machine motion state at the current moment;

in the formula, x_div(k + 1) is a drive state vector of the drive state observation layer at the current time, x_div(k) Observing the driving state vector of the layer for the driving state of the previous moment, Q_div(k) Covariance of process noise, y, for the observation layer of the drive state at the previous time_DMFSTF(k + 1) is the measurement vector of the observer of the state of motion at the current moment, H_div(k + 1) is the measured Jacobian matrix of the driving observation model at the current moment, R_vicAnd (k + 1) is the covariance of the measurement noise of the observation layer of the complete machine motion state at the current moment.

Step S20: calculating a complete machine prior estimation value corresponding to the complete machine motion state observation layer at the current moment according to the complete machine observation model, and calculating a driving prior estimation value corresponding to the driving state observation layer at the current moment according to the driving observation model;

in a specific implementation, according to the complete machine observation model, a complete machine prior estimation value corresponding to the complete machine motion state observation layer at the current moment can be calculated through the following formula,

in the formula (I), the compound is shown in the specification,

according to the driving observation model, a driving prior estimation value corresponding to the driving state observation layer at the current moment can be calculated through the following formula,

in the formula (I), the compound is shown in the specification,

Step S30: respectively calculating a complete machine multiple time-varying fading factor matrix corresponding to the complete machine motion state observation layer and a driving multiple time-varying fading factor matrix corresponding to the driving state observation layer;

in a specific implementation, a complete machine multiple time-varying fading factor matrix corresponding to the complete machine motion state observation layer can be calculated through the following formula,

in the formula, λ_divFor driving multiple time-varying fading factor matrix, lambda, corresponding to the driving state observation layer_{div_T_lf}Multiple time-varying fading factors, lambda, corresponding to the driving torque of the hub motor of the left front wheel_{div_T_rf}Multiple time-varying fading factors, lambda, corresponding to the driving torque of the hub motor of the right front wheel_{div_T_lr}Multiple time-varying fading factors, lambda, corresponding to the driving torque of the hub motor of the left rear wheel_{div_T_rr}Multiple time-varying fading factors corresponding to the driving torque of the hub motor of the right rear wheel, T_{_lf}Driving torque, T, of a hub motor for the left front wheel_{_rf}Driving torque, T, of a hub motor for the right front wheel_{_lr}Driving torque, T, of a hub motor for a left rear wheel_{_rr}Driving torque, lambda, of a hub motor for the right rear wheel_{div_i}(k + 1) is the current timeMultiple time-varying fading factors, η, of the drives corresponding to the observed layer of states of the drives_{div_i}And (k + 1) is a preset multiple parameter of the driving state observation layer at the current moment.

Wherein the preset multi-parameter matrix of the whole machine motion state observation layer at the current moment is obtained by the following formula,

in the formula eta_vicFor presetting multiple parameter matrixes, eta, of the observation layer of the motion state of the whole machine_{vic_1}、η_{vic_2}……η_{vic_n}Different multi-parameter matrices, eta, for the observation layer of the state of motion of the machine_vic(k + 1) is a preset multi-parameter matrix of the driving state observation layer at the current moment, F_vic(k) A system state Jacobian matrix H corresponding to the linearized whole machine observation model at the previous moment_vic(k + 1) is a measured Jacobian matrix, Q, corresponding to the linearized whole machine observation model at the current moment_vic(k) Covariance of process noise, beta, of the observation layer of the complete machine motion state at the previous moment_vicMultiple weakening factors, beta, for the observation layer of the state of motion of the whole machine_{vic_1}、β_{vic_2}……β_{vic_n}For different multiple weakening factors, R, of the observation layer of the state of motion of the machine_vic(k + 1) is the covariance of the measurement noise of the observation layer of the complete machine motion state at the current moment, P_vic(k) Forecasting the variance matrix, gamma, for the whole machine prior corresponding to the observation layer of the whole machine motion state at the previous moment_vicIs the residual error of the observation layer of the motion state of the whole machine, V_{0_vic}(k + 1) is a residual covariance matrix of the observation layer of the complete machine motion state at the current moment, V_{0_vic}(k) Is a residual covariance matrix, rho, of the observation layer of the complete machine motion state at the previous moment_vicMultiple observation layers for the motion state of the whole machineForgetting factor, ρ_{vic_1}、ρ_{vic_2}……ρ_{vic_n}Different multiple forgetting factors, gamma, for the observation layer of the complete machine motion state_vic(k + 1) is the residual error of the observation layer of the complete machine motion state at the current moment;

in the formula eta_divPredetermined multiple parameter matrix, η, for driving the state observation layers_{div_T_lf}Is a multi-parameter matrix, eta, corresponding to the driving torque of the hub motor of the left front wheel_{div_T_rf}A multi-parameter matrix, eta, corresponding to the driving torque of the hub motor of the right front wheel_{div_T_lr}A multi-parameter matrix, eta, corresponding to the driving torque of the hub motor of the left rear wheel_{div_T_rr}A matrix of multiple parameters, eta, corresponding to the driving torque of the hub motor of the right rear wheel_div(k + 1) is a preset multi-parameter matrix of the driving state observation layer at the current moment, F_div(k) The Jacobian matrix of the system states, H, corresponding to the driving observation model at the previous moment_div(k + 1) is a measured Jacobian matrix corresponding to the driving observation model at the current moment, V_{0_div}(k) Residual covariance matrix, V, for the observation layer of the drive state at the previous moment_{0_div}(k + 1) is the residual covariance matrix of the observation layer of the drive state at the previous time, Q_div(k) Covariance of process noise, beta, for the observation layer of the driving state at the previous moment_divMultiple attenuation factors, β, for driving the state observation layers_{div_lf}Multiple attenuation factors, beta, corresponding to the left front wheel in the drive state observation horizon_{div_rf}Multiple attenuation factors, beta, corresponding to the right front wheel in the driving state observation layer_{div_lr}Multiple attenuation factors, beta, corresponding to the left rear wheel in the drive state observation layer_{div_rr}Multiple attenuation factors, R, corresponding to the right rear wheel in the drive status observation layer_vic(k + 1) is the covariance of the measurement noise of the observation layer of the complete machine motion state at the current moment, P_div(k) Predicting a variance matrix, gamma, for a drive prior corresponding to the observation layer of the drive state at the previous moment_divFor the residual of the observation layer in the driven state, p_divMultiple forgetting factor, rho, for driving the state observation layer_{div_lf}Multiple forgetting factors, rho, corresponding to the left front wheel in the observation layer of the driving state_{div_rf}Multiple forgetting factors, rho, corresponding to the right front wheel in the drive state observation layer_{div_lr}Multiple forgetting factors, rho, corresponding to the left rear wheel in the drive state observation layer_{div_rr}Multiple forgetting factors, gamma, corresponding to the right rear wheel in the drive state observation layer_div(k + 1) is the residual of the drive state observation layer at the current time.

Wherein k may be understood as a previous time, and k +1 may be understood as a current time (which is not limited herein, and only represents a temporal precedence relationship, and may also be understood as k being the current time, and k +1 being the next time); n is the number of observed quantities of the complete machine motion state, and can be understood as the dimensionality of the complete machine motion state vector. Matrix element value ranges of multiple weakening factors of the whole machine motion state observation layer and the driving state observation layer can be 0-20, and different curve smoothness can be obtained through different values; matrix element value ranges of multiple weakening factors of the complete machine motion state observation layer and the driving state observation layer can be 0.5-1, and different values can weaken the influence of data at the previous moment on data observation at the current moment to different degrees.

Step S40: calculating a complete machine prior forecast variance matrix corresponding to the complete machine motion state observation layer according to the complete machine multiple time-varying fading factor matrix and the improved complete machine covariance matrix, and calculating a drive prior forecast variance matrix corresponding to the drive motion state observation layer according to the drive multiple time-varying fading factor matrix and the improved drive covariance matrix;

in a specific implementation, a complete machine priori prediction variance matrix corresponding to the complete machine motion state observation layer can be calculated according to the complete machine multiple time-varying fading factor matrix and the improved complete machine covariance matrix through the following formula,

in the formula, P_div(k +1 | k) is a drive prior prediction variance matrix corresponding to the drive state observation layer at the current time predicted from the drive prior prediction variance matrix corresponding to the drive state observation layer at the previous time, P_div(k) Predicting a variance matrix, lambda, for a drive prior corresponding to the observation layer of the drive state at the previous time_div(k + 1) is a driving multiple time-varying fading factor matrix corresponding to the driving state observation layer at the current moment,Q_div(k) the covariance of the process noise of the layer is observed for the drive state at the previous time instant.

Step S50: calculating a complete machine state correction gain corresponding to the complete machine motion state observation layer according to the complete machine priori prediction variance matrix, and calculating a driving state correction gain corresponding to the driving state observation layer according to the driving priori prediction variance matrix;

in a specific implementation, the total machine state correction gain corresponding to the observation layer of the total machine motion state can be calculated according to the total machine priori forecast variance matrix through the following formula,

the driving state correction gain corresponding to the driving state observation layer can be calculated according to the driving prior forecast variance matrix through the following formula,

in the formula, K_div(k + 1) is a drive state correction gain corresponding to the drive state observation layer, P_div(k +1 | k) is a drive prior prediction variance matrix corresponding to the drive state observation layer at the current time predicted from the drive prior prediction variance matrix corresponding to the drive state observation layer at the previous time, H_div(k + 1) is the quantity of the driving observation model at the current timeJacobian matrix, R_vic(k) The covariance of the measurement noise of the observation layer of the complete machine motion state at the previous moment is obtained.

Step S60: correcting the complete machine prior estimated value according to the complete machine state correction gain to obtain a complete machine optimal estimated value, and correcting the drive prior estimated value according to the drive state correction gain to obtain a drive optimal estimated value;

in a specific implementation, the complete machine prior estimated value can be corrected according to the complete machine state correction gain through the following formula to obtain a complete machine optimal estimated value,

in the formula (I), the compound is shown in the specification,

the driving prior estimated value can be corrected according to the driving state correction gain through the following formula to obtain a driving optimal estimated value,

in the formula (I), the compound is shown in the specification,

(K +1 | K) is a drive prior estimate value corresponding to the current-time drive state observation layer predicted from the drive prior estimate value corresponding to the previous-time drive state observation layer, and K_div(k + 1) is a drive state correction gain corresponding to the drive state observation layer, y_DMFSTF(k + 1) is the measurement vector of the observer of the state of motion at the current moment, H_divAnd (k + 1) is a measured Jacobian matrix of the driving observation model at the current moment.

Step S70: and observing the motion state at the next moment based on the complete machine optimal estimation value, the drive optimal estimation value, the complete machine prior prediction variance matrix and the drive prior prediction variance matrix.

In a specific implementation, after obtaining a total machine prior prediction variance matrix corresponding to the observation layer of the total machine motion state, the method further includes:

in the formula, P_vic(k + 1) is the optimal variance matrix of the whole machine corresponding to the observation layer of the whole machine motion state at the current moment, F_vic(k) A system state Jacobian matrix, P, corresponding to the linearized whole machine observation model at the previous moment_vic(k +1 | k) is a total machine prior prediction variance matrix corresponding to the total machine motion state observation layer at the current time predicted according to the total machine prior prediction variance matrix corresponding to the total machine motion state observation layer at the previous time, and Q is_vic（k) The covariance of the process noise of the whole machine motion state observation layer at the previous moment,

_{vic_1}、

_{vic_2}、……

_{div_lf}for driving state observation layer left front wheelPositive definite parameters,

Further, the motion state observation at the next moment can be performed based on the complete machine optimal variance matrix obtained by correcting the complete machine priori prediction variance matrix, the drive optimal variance matrix obtained by correcting the drive priori prediction variance matrix, the complete machine optimal estimation value and the drive optimal estimation value, so that the accuracy of the complete machine motion state observation under the drive fault is ensured.

The self-adaptive precise identification method is easy to understand, different time-varying fading factors are set for each driving state and the complete machine motion state, so that multiple observation requirements of motion control on different states are met, the problem of information divergence of the motion state of the distributed electric chassis caused by the introduction of multiple time-varying fading factors is improved by introducing fixed parameters, and further, the self-adaptive precise identification of different motion states of the distributed electric chassis is realized by combining strong tracking of the two observation layer mutation states and data interaction between the observation layers.

In the embodiment, a double-structure multiple fading factor strong tracking motion state observer is designed for the motion state observation of the distributed electric chassis under the drive fault, a drive state observation layer and a complete machine motion state observation layer are respectively designed based on a multiple fading factor strong tracking filter, different time-varying fading factors are respectively introduced for each drive state and complete machine motion state, so that the observation requirements of different motion states are met, and the fast and accurate tracking of the motion states of different distributed electric chassis under the drive fault is realized by combining the strong tracking of the states of the two observation layers and the data interaction between the two observation layers. Furthermore, in order to avoid the influence of positive qualitative damage caused by introducing multiple fading factors on the overall motion state observation precision of the distributed electric chassis, positive parameters are respectively introduced into the driving state observation layer and the overall motion state observation layer to improve the driving state observation layer and the overall motion state observation layer, so that the accuracy of motion state observation under the driving fault is further ensured.

Referring to fig. 2, fig. 2 is a block diagram illustrating a first embodiment of a motion state observation system based on a distributed electric chassis according to the present invention.

As shown in fig. 2, a motion state observation system based on a distributed electric chassis according to an embodiment of the present invention includes:

the model building module 10 is used for respectively building a complete machine observation model corresponding to a complete machine motion state observation layer and a driving observation model corresponding to a driving state observation layer of the motion state observer based on the distributed electric chassis;

a priori value calculation module 20, configured to calculate a complete machine priori estimation value corresponding to the complete machine motion state observation layer at the current time according to the complete machine observation model, and calculate a drive priori estimation value corresponding to the drive state observation layer at the current time according to the drive observation model;

a factor calculating module 30, configured to calculate a complete machine multiple time-varying fading factor matrix corresponding to the complete machine motion state observation layer and a driving multiple time-varying fading factor matrix corresponding to the driving state observation layer respectively;

a variance calculation module 40, configured to calculate a complete machine priori prediction variance matrix corresponding to the complete machine motion state observation layer according to the complete machine multiple time-varying fading factor matrix and the improved complete machine covariance matrix, and calculate a drive priori prediction variance matrix corresponding to the drive motion state observation layer according to the drive multiple time-varying fading factor matrix and the improved drive covariance matrix;

a correction module 50, configured to calculate a complete machine state correction gain corresponding to the complete machine motion state observation layer according to the complete machine priori prediction variance matrix, and calculate a driving state correction gain corresponding to the driving state observation layer according to the driving priori prediction variance matrix;

an optimal value calculation module 60, configured to modify the complete machine priori estimated value according to the complete machine state modification gain to obtain a complete machine optimal estimated value, and modify the drive priori estimated value according to the drive state modification gain to obtain a drive optimal estimated value;

and a recursive observation module 70, configured to perform motion state observation at the next time based on the complete machine optimal estimated value, the drive optimal estimated value, the complete machine prior prediction variance matrix, and the drive prior prediction variance matrix.

Other embodiments or specific implementation manners of the motion state observation system based on the distributed electric chassis can refer to the above method embodiments, and are not described herein again.

It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or system. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or system that comprises the element.

The above-mentioned serial numbers of the embodiments of the present invention are merely for description and do not represent the merits of the embodiments.

Through the above description of the embodiments, those skilled in the art will clearly understand that the method of the above embodiments can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware, but in many cases, the former is a better implementation manner. Based on such understanding, the technical solutions of the present invention may be embodied in the form of a software product, which is stored in a storage medium (e.g., a rom/ram, a magnetic disk, an optical disk) and includes instructions for enabling a terminal device (e.g., a mobile phone, a computer, a server, an air conditioner, or a network device) to execute the method according to the embodiments of the present invention.

The above description is only a preferred embodiment of the present invention, and not intended to limit the scope of the present invention, and all modifications of equivalent structures and equivalent processes, which are made by using the contents of the present specification and the accompanying drawings, or directly or indirectly applied to other related technical fields, are included in the scope of the present invention.

Claims

1. A motion state observation method based on a distributed electric chassis is characterized by comprising the following steps:

observing the motion state at the next moment based on the complete machine optimal estimated value, the drive optimal estimated value, the complete machine priori prediction variance matrix and the drive priori prediction variance matrix;

the step of respectively constructing a complete machine observation model corresponding to a complete machine motion state observation layer and a drive observation model corresponding to a drive state observation layer in the motion state observer based on the distributed electric chassis specifically includes:

in the formula, x_div(k + 1) is a drive state vector of the drive state observation layer at the current time, x_div(k) Observing the driving state vector of the layer for the driving state of the previous moment, Q_div(k) Covariance of process noise of observation layer for driving state of previous moment, H_div(k + 1) is a measured Jacobian matrix of the driving observation model at the current moment;

the step of calculating a complete machine prior estimation value corresponding to the complete machine motion state observation layer at the current moment according to the complete machine observation model and calculating a drive prior estimation value corresponding to the drive state observation layer at the current moment according to the drive observation model specifically includes:

in the formula (I), the compound is shown in the specification,

in the formula (I), the compound is shown in the specification,

(k) the driving prior estimation value is corresponding to the driving state observation layer at the previous moment;

the step of calculating the complete machine multiple time-varying fading factor matrix corresponding to the complete machine motion state observation layer and the driving multiple time-varying fading factor matrix corresponding to the driving state observation layer respectively specifically includes:

in the formula, λ_vicIs a complete machine multiple time-varying fading factor matrix, lambda, corresponding to the observation layer of the complete machine motion state_{vic_1}、λ_{vic_2}……λ_{vic_n}Multiple time-varying fading factors, x, corresponding to different overall motion states in the overall motion state observation layer_vic（1）、x_vic（1）……x_vic(n) different overall machine motion state vectors of the overall machine motion state observation layer,λ_{vic_i}(k + 1) is a multiple time-varying fading factor, eta, of the whole machine corresponding to the observation layer of the whole machine motion state at the current moment_{vic_i}(k + 1) is a preset multi-parameter matrix of the observation layer of the complete machine motion state at the current moment;

in the formula, λ_divFor driving multiple time-varying fading factor matrix, lambda, corresponding to the driving state observation layer_{div_T_lf}Multiple time-varying fading factors, lambda, corresponding to the driving torque of the hub motor of the left front wheel_{div_T_rf}Multiple time-varying fading factors, lambda, corresponding to the driving torque of the hub motor of the right front wheel_{div_T_lr}Multiple time-varying fading factors, lambda, corresponding to the driving torque of the hub motor of the left rear wheel_{div_T_rr}Multiple time-varying fading factors corresponding to the driving torque of the hub motor of the right rear wheel, T_{_lf}Driving torque, T, of a hub motor for the left front wheel_{_rf}Driving torque, T, of a hub motor for the right front wheel_{_lr}Driving torque, T, of a hub motor for a left rear wheel_{_rr}Driving torque, lambda, of a hub motor for the right rear wheel_{div_i}(k + 1) is a driving multiple time-varying fading factor, η, corresponding to the driving state observation layer at the current time_{div_i}(k + 1) is a preset multiple parameter matrix of the driving state observation layer at the current moment;

the preset multi-parameter matrix of the whole machine motion state observation layer at the current moment is obtained by the following formula,

in the formula eta_vicFor presetting multiple parameter matrixes, eta, of the observation layer of the motion state of the whole machine_{vic_1}、η_{vic_2}……η_{vic_n}For observing the motion state of the whole machineDifferent multiple parameter matrices, η_vic(k + 1) is a preset multiple parameter matrix of the observation layer of the complete machine motion state at the current moment, F_vic(k) A system state Jacobian matrix H corresponding to the linearized whole machine observation model at the previous moment_vic(k + 1) is a measured Jacobian matrix, Q, corresponding to the linearized whole machine observation model at the current moment_vic(k) Covariance of process noise, beta, of the observation layer of the complete machine motion state at the previous moment_vicMultiple weakening factors, R, for the observation layer of the state of motion of the whole machine_vic(k + 1) is the covariance of the measurement noise of the observation layer of the complete machine motion state at the current moment, P_vic(k) Forecasting the variance matrix, gamma, for the whole machine prior corresponding to the observation layer of the whole machine motion state at the previous moment_vicIs the residual error of the observation layer of the motion state of the whole machine, V_{0_vic}(k + 1) is a residual covariance matrix of the observation layer of the complete machine motion state at the current moment, V_{0_vic}(k) Is a residual covariance matrix, rho, of the observation layer of the complete machine motion state at the previous moment_vicMultiple forgetting factor, gamma, of the observation layer of the motion state of the whole machine_vic(k + 1) is the residual error of the observation layer of the complete machine motion state at the current moment;

in the formula eta_divPredetermined multiple parameter matrix, η, for driving the state observation layers_{div_T_lf}Is a multi-parameter matrix, eta, corresponding to the driving torque of the hub motor of the left front wheel_{div_T_rf}A multi-parameter matrix, eta, corresponding to the driving torque of the hub motor of the right front wheel_{div_T_lr}The driving torque of the hub motor of the left rear wheel corresponds to moreThe weight parameter matrix, η_{div_T_rr}A matrix of multiple parameters, eta, corresponding to the driving torque of the hub motor of the right rear wheel_div(k + 1) is a preset multi-parameter matrix of the driving state observation layer at the current moment, F_div(k) The Jacobian matrix of the system states, H, corresponding to the driving observation model at the previous moment_div(k + 1) is a measured Jacobian matrix corresponding to the driving observation model at the current moment, V_{0_div}(k) Residual covariance matrix, V, for the observation layer of the drive state at the previous moment_{0_div}(k + 1) is the residual covariance matrix of the observation layer of the drive state at the previous time, Q_div(k) Covariance of process noise, beta, for the observation layer of the driving state at the previous moment_divMultiple attenuation factors, P, for driving the state observation layers_div(k) Predicting a variance matrix, gamma, for a drive prior corresponding to the observation layer of the drive state at the previous moment_divFor the residual of the observation layer in the driven state, p_divMultiple forgetting factor, gamma, for driving the state observation layer_div(k + 1) is the residual error of the driving state observation layer at the current moment;

the step of calculating a complete machine priori prediction variance matrix corresponding to the complete machine motion state observation layer according to the complete machine multiple time-varying fading factor matrix and the improved complete machine covariance matrix, and calculating a drive priori prediction variance matrix corresponding to the drive motion state observation layer according to the drive multiple time-varying fading factor matrix and the improved drive covariance matrix specifically includes:

in the formula, P_vic(k +1 | k) is the total motion state observation at the current time predicted according to the total machine priori prediction variance matrix corresponding to the total motion state observation layer at the previous timeLayer-wise ensemble prior prediction variance matrix, P_vic(k) Forecasting the variance matrix, lambda, of the whole machine prior corresponding to the observation layer of the whole machine motion state at the previous moment_vic(k + 1) is a complete machine multiple time-varying fading factor matrix corresponding to the complete machine motion state observation layer at the current moment, F_vic(k) A system state Jacobian matrix, Q, corresponding to the linearized whole machine observation model at the previous moment_vic(k) The covariance of the process noise of the whole machine motion state observation layer at the previous moment is obtained;

in the formula, P_div(k +1 | k) is a drive prior prediction variance matrix corresponding to the drive state observation layer at the current time predicted from the drive prior prediction variance matrix corresponding to the drive state observation layer at the previous time, P_div(k) Predicting a variance matrix, lambda, for a drive prior corresponding to the observation layer of the drive state at the previous time_div(k + 1) is a driving multiple time-varying fading factor matrix, Q, corresponding to the driving state observation layer at the current time_div(k) The covariance of the process noise of the observation layer for the drive state at the previous time;

the step of calculating the complete machine state correction gain corresponding to the complete machine motion state observation layer according to the complete machine priori prediction variance matrix and calculating the driving state correction gain corresponding to the driving state observation layer according to the driving priori prediction variance matrix specifically includes:

in the formula, K_div(k + 1) is a drive state correction gain corresponding to the drive state observation layer, P_div(k +1 | k) is a drive prior prediction variance matrix corresponding to the drive state observation layer at the current time predicted from the drive prior prediction variance matrix corresponding to the drive state observation layer at the previous time, H_div(k + 1) is a measured Jacobian matrix of the driving observation model at the current moment;

the step of correcting the complete machine prior estimated value according to the complete machine state correction gain to obtain a complete machine optimal estimated value, and correcting the drive prior estimated value according to the drive state correction gain to obtain a drive optimal estimated value specifically includes:

in the formula (I), the compound is shown in the specification,

in the formula (I), the compound is shown in the specification,

(K +1 | K) is a drive prior estimate value corresponding to the current-time drive state observation layer predicted from the drive prior estimate value corresponding to the previous-time drive state observation layer, and K_div(k + 1) is a drive state correction gain corresponding to the drive state observation layer, H_div(k + 1) is a measured Jacobian matrix of the driving observation model at the current moment;

after obtaining the complete machine prior forecast variance matrix corresponding to the complete machine motion state observation layer, the method further comprises the following steps:

in the formula, P_vic(k + 1) is the optimal variance matrix of the whole machine corresponding to the observation layer of the whole machine motion state at the current moment, F_vic(k) A system state Jacobian matrix, P, corresponding to the linearized whole machine observation model at the previous moment_vic(k +1 | k) is a total machine prior prediction variance matrix corresponding to the total machine motion state observation layer at the current time predicted according to the total machine prior prediction variance matrix corresponding to the total machine motion state observation layer at the previous time, and Q is_vic(k) The covariance of the process noise of the whole machine motion state observation layer at the previous moment,

_{vic_1}、

_{vic_2}、……

_{div_rr}Positive definite parameter, K, corresponding to the right rear wheel in the observation layer of driving state_div(k + 1) is a drive state correction gain corresponding to the drive state observation layer, H_div(k + 1) is a measured Jacobian matrix corresponding to the driving observation model at the current moment, lambda_div(k + 1) is the multiple time-varying fading of the drive corresponding to the observation layer of the driving state at the current timeA factor matrix.

2. A distributed electric chassis-based motion state observation system for performing the distributed electric chassis-based motion state observation method according to claim 1, the distributed electric chassis-based motion state observation system comprising: