CN116552548A - Four-wheel distributed electric drive automobile state estimation method - Google Patents

Abstract

The invention provides a state estimation method of a four-wheel distributed electric drive automobile, which takes the motion of the four-wheel distributed electric drive automobile into consideration, establishes a three-degree-of-freedom vehicle dynamics model, takes the running parameters fed back by the four-wheel distributed electric drive automobile in real time as model input quantity, utilizes system input parameters, vehicle structure parameters and real-time measurement parameters, fully takes the characteristic of real-time feedback of the vehicle parameters of the distributed electric drive automobile and the optimizing capability of non-Gaussian noise of the system into consideration, and simultaneously takes the longitudinal motion, transverse motion and yaw motion of the vehicle into consideration, so that the running characteristics of the distributed electric drive automobile under various working conditions can be reflected more accurately.

Description

A state estimation method for four-wheel distributed electric drive vehicles

technical field

The invention relates to the technical field of electric vehicles, in particular to a state estimation method for a four-wheel distributed electric drive vehicle.

Background technique

Electric vehicles have the advantages of energy saving and environmental protection, and are an important means to realize the energy transformation of vehicles. Four-wheel distributed electric drive vehicle is the main type of electric vehicle development. Compared with traditional vehicles, its braking torque and driving torque can be independently controlled, its power transmission efficiency is high, and it is easier to realize vehicle active safety control, which is the development direction of future automobiles.

However, the premise of the control and decision-making of the vehicle active safety control system is to accurately obtain the vehicle state parameter information. Due to technical limitations or the high price of some sensors, or the signal is greatly affected by external interference, it is difficult to measure directly. Therefore, how to accurately estimate these difficult-to-measure state parameters, such as the center-of-mass slip angle, based on some existing low-cost sensors is a technical problem to be solved by those skilled in the art.

Contents of the invention

The purpose of the present invention is to provide a method for estimating the state of a four-wheel distributed electric drive vehicle, so as to realize effective real-time estimation of the state of the four-wheel distributed electric drive vehicle.

A method for state estimation of a four-wheel distributed electric drive vehicle, comprising the following steps:

Step 1. According to the motion characteristics of the distributed electric drive vehicle, construct a longitudinal, lateral, and yaw nonlinear three-degree-of-freedom vehicle dynamics model, and construct a state-space equation based on the nonlinear three-degree-of-freedom vehicle model;

Step 2, based on the state space equation constructed in step 1, connect the square root volumetric Kalman filter with the vehicle dynamics system, establish the discrete state equation and observation equation of the square root volumetric Kalman filter of the nonlinear system, and determine the square root volumetric Kalman filter The estimated quantity, input quantity and observed quantity of the Mann filter, and the square root volumetric Kalman filter is used to correct and recursively update the estimated parameters of the square root volumetric Kalman filter;

Step 3, based on the square root volumetric Kalman filter constructed in step 2, using the maximum correlation entropy criterion as an optimization standard to construct a maximum correlation entropy square root volumetric Kalman filter;

Step 4, obtain the improved African vulture algorithm, and optimize the non-Gaussian noise of the maximum correlation entropy square root volumetric Kalman filter based on step 3, and optimize the maximum correlation entropy square root volumetric Kalman filter through the improved African vulture algorithm Finally, the optimal estimation of the state parameters of the four-wheel distributed electric drive vehicle is realized.

According to the method for estimating the state of a four-wheel distributed electric drive vehicle provided by the present invention, it has the following beneficial effects:

1. The present invention considers the motion of the four-wheel distributed electric drive vehicle, establishes a three-degree-of-freedom vehicle dynamics model, and takes the driving parameters (wheel rotation angle and lateral acceleration) fed back in real time by the four-wheel distributed electric drive vehicle as model input, Using the system input parameters, vehicle structure parameters, and real-time measurement parameters, an improved African vulture algorithm is established to optimize the maximum correlation entropy square root volumetric Kalman filter to estimate the key state parameters of the handling stability of the four-wheel distributed electric drive vehicle in real time;

2. Different from the general online estimation method of vehicle state parameters, the present invention fully considers the characteristics of real-time feedback of vehicle parameters of distributed drive electric vehicles and the ability to optimize the non-Gaussian noise of the system, and at the same time, the model considers the longitudinal and lateral movements of the vehicle and yaw motion, which can accurately reflect the driving characteristics of distributed drive electric vehicles under various working conditions;

3. Different from the general unscented Kalman filter algorithm and the extended Kalman filter algorithm, the present invention uses the improved African vulture optimized maximum correlation entropy square root volumetric Kalman filter to estimate the state of the established vehicle dynamics system model. Considering the dynamic model of distributed electric drive vehicles driving under common road conditions is a nonlinear system, the traditional unscented Kalman filter algorithm and extended Kalman filter algorithm have low estimation accuracy, while the volumetric Kalman filter algorithm uses a third-order sphere The radial volume criterion, and using a set of volume points to approximate the state mean and covariance of a nonlinear system with additional Gaussian noise, is currently the closest approximation algorithm to Bayesian filtering in theory, and is a solution to high-order nonlinear system states A powerful tool for estimation. However, in the recursive process of the volumetric Kalman filter, the calculation amount is large and the value is unstable. Therefore, on the basis of the volumetric Kalman filter, the present invention introduces the idea of the square root filter and uses the maximum correlation entropy criterion to constrain the covariance matrix of the process noise, and adopts the improved African vulture algorithm to the maximum correlation entropy square root volumetric Kalman filter. Adaptive optimization of non-Gaussian noise can reduce the influence of unstable filtering process to a certain extent and improve the estimation accuracy;

4. The vehicle state parameter estimation algorithm based on the dynamic model involved in the present invention combines the characteristics that the vehicle information parameters of the four-wheel distributed electric drive vehicle can be obtained in real time, which reduces the requirements for the on-board sensors. At the same time, the estimation process is basically not affected by changes in vehicle structural parameters such as vehicle mass and vehicle yaw moment of inertia, and has wide applicability and good robustness.

Description of drawings

Fig. 1 is the flow chart of the four-wheel distributed electric drive vehicle state estimation method provided by the embodiment of the present invention;

Fig. 2 is a comparison diagram between the method proposed by the present invention and the traditional volumetric Kalman algorithm and the actual value.

Detailed ways

In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments It is a part of embodiments of the present invention, but not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the protection scope of the present invention.

Please refer to Fig. 1, an embodiment of the present invention provides a method for estimating the state of a four-wheel distributed electric drive vehicle. In this embodiment, the state of a four-wheel distributed electric drive vehicle is described by taking the side slip angle of the center of mass as an example. The method includes Step 1~Step 4:

Step 1. According to the motion characteristics of the distributed electric drive vehicle, construct a longitudinal, lateral, and yaw nonlinear three-degree-of-freedom vehicle dynamics model, and construct a state-space equation based on the nonlinear three-degree-of-freedom vehicle model.

In step 1, the nonlinear three-degree-of-freedom vehicle dynamics equation established according to the longitudinal motion, lateral motion, and yaw motion of the distributed electric drive vehicle is shown in the following formula:

in, , /> are vehicle yaw rate and side slip angle of center of mass respectively; /> represents the front wheel rotation angle; v _x , a _x , a _y represent the longitudinal velocity, longitudinal acceleration and lateral acceleration of the vehicle respectively; k ₁ , k ₂ represent the equivalent cornering stiffness of the front axle and rear axle of the vehicle respectively; is the distance from the front axle to the center of mass, and the distance from the rear axle to the center of mass; I _z is the moment of inertia of the center of mass; m is the mass of the vehicle; /> expresses the differentiation of the indicated quantity;

The state space equation of the established nonlinear three-degree-of-freedom vehicle model is as follows:

.

Step 2, based on the state space equation constructed in step 1, connect the square root volumetric Kalman filter with the vehicle dynamics system, establish the discrete state equation and observation equation of the square root volumetric Kalman filter of the nonlinear system, and determine the square root volumetric Kalman filter The estimated quantity, input quantity and observed quantity of the Mann filter, and the square root volumetric Kalman filter is used to correct and recursively update the estimated parameters of the square root volumetric Kalman filter.

Among them, step 2 specifically includes:

The discrete state equation and observation equation of the square root volumetric Kalman filter to construct the nonlinear system are as follows:

in, is the state variable at time k , /> , T represents the transpose operation; /> is the control variable at time k , /> ;/> is the observed variable at time k , /> ;/> is the state variable at time k -1; /> is the control variable at time k -1; f is the evolution equation of the discrete state equation; h is the evolution equation of the observation equation, /> and are uncorrelated process noise and measurement noise with zero mean, respectively;

The forward Euler discretization method is used to obtain the discretized state equation and measurement equation, as shown in the following formula:

in, , /> are respectively the vehicle yaw rate and the side slip angle of the center of mass at time k ; /> , /> Respectively represent the longitudinal velocity and lateral acceleration of the vehicle at time k ; /> , /> are the vehicle yaw rate and the side slip angle of the center of mass at time k -1 respectively; /> , /> , /> Respectively represent the longitudinal velocity, longitudinal acceleration and lateral acceleration of the vehicle at time k -1; Indicates the front wheel angle of the vehicle at time k -1;

The discretized state equation and measurement equation are introduced into the square root volumetric Kalman filter for prediction and update.

The discretized state equation and measurement equation are introduced into the square root volumetric Kalman filter, and the prediction and update specifically include:

1) Forecast:

At time k , define is the state covariance matrix at time k -1 /> the square root of is the covariance of the process noise /> square root factor of , /> is the covariance of the measurement noise /> square root factor of

Calculation of state volume points:

in, is the state volume point, /> is the i-th volume point, /> , n is a natural number greater than or equal to 1, /> is the estimated state value at time k -1;

Propagation of state volume points:

in, is the predicted value corresponding to the volume point in the prediction process;

The prior state and the square root of the covariance matrix are estimated by:

in, is the state variable at time k predicted according to time k -1;

in, is the square root of the state covariance matrix at k time predicted according to k -1 time, /> represents the triangular decomposition of a matrix, /> is the deviation matrix of the state vector;

2) Update:

Calculation of measuring volume points:

in, The state vector corresponding to the volume point reconstructed for the update phase, /> is the state vector at time k estimated according to time k -1 in the update phase, /> is the square root of the state covariance matrix at time k -1 of the update reconstruction stage;

Measure the spread of volume points:

in, is the predicted value corresponding to the volume point during the measurement update process;

The a priori measurements and the square root of the covariance matrix are estimated by:

in, is the measured variable at time k predicted according to time k -1;

in, is the measurement covariance matrix at k time predicted according to k -1 time, /> is the deviation matrix of the measurement vector;

Compute the cross-covariance matrix:

in, is the transpose matrix of the bias matrix of the measurement vector, /> is the cross-covariance matrix, is the deviation matrix of the state vector;

Calculate the Kalman gain :

in, is the square root factor of the innovation variance, /> is the transposed matrix of the square root factor of the innovation variance;

The posterior state and the square root of the covariance matrix are estimated by:

in, is the expected state variable at time k , /> is the measured variable collected at time k ;

in, is the square root of the state covariance matrix at time k .

Step 3. Based on the square root volumetric Kalman filter constructed in step 2, the maximum correlation entropy criterion is used as the optimization standard to construct a maximum correlation entropy square root volumetric Kalman filter.

Using the maximum correlation entropy criterion as the optimization standard, the maximum correlation entropy square root volumetric Kalman filter is constructed, which fully considers the high-order moments of the estimation error, and shows strong robustness when dealing with non-Gaussian noise.

Among them, step 3 specifically includes:

Combining the nonlinear three-degree-of-freedom vehicle dynamics model constructed in step 1 with the square root volumetric Kalman filter constructed in step 2, the maximum correlation entropy criterion is introduced to construct a nonlinear recursive model:

in, for the process variable, /> , /> covariance of for:

in, is the process variable corresponding to the state error covariance, /> is the process variable corresponding to the measurement error covariance, /> is the state covariance matrix at time k ;

Then the best estimate based on the maximum correlation entropy criterion is obtained as follows:

in, is the state quantity best estimated by the maximum correlation entropy criterion, /> is the kernel function, q + m is the dimension;

Then, define is the process variable of the maximum correlation entropy criterion, and the following formula is obtained:

in, for /> Diagonal matrix of the first q elements, /> for /> The diagonal matrix of the q +1 to q + m elements of ;

Will The updated covariance matrix of is defined as:

in, is the state covariance matrix at time k after the maximum correlation entropy improvement, /> The measurement error covariance matrix improved by the maximum correlation entropy; /> for /> the inverse matrix;

The prior measurement noise variance can be expressed as:

in, for /> inverse matrix;

Then, the square root of the updated measurement covariance matrix Obtained by the following formula:

Finally, the into the square root volumetric Kalman filter instead of /> An iterative cycle is performed to construct a maximum correlation entropy square root volumetric Kalman filter.

Step 4, obtain the improved African vulture algorithm, and optimize the non-Gaussian noise of the maximum correlation entropy square root volumetric Kalman filter based on step 3, and optimize the maximum correlation entropy square root volumetric Kalman filter through the improved African vulture algorithm Finally, the optimal estimation of the state parameters of the four-wheel distributed electric drive vehicle is realized.

In step 4, the following improvements are made to the African vulture algorithm:

The first stage: determine the best vulture in the group

The chaotic optimization mode is introduced into the African condor optimization algorithm, and the principle of chaos theory is used to explore the search space. The expression is as follows:

in, is the global best position at the current moment, /> For the global best position at the next moment; introduce chaos optimization mode, use the principle of chaos theory to explore the search space, make the population initialization more uniform, and avoid local optimum.

After initializing the population, calculate the population fitness value:

in, is the probability that other vultures move to the optimal vulture position, /> and /> is the parameter given before the search operation, with a value between 0 and 1, and /> and /> The sum is 1; /> , /> Two groups of best vultures respectively;/> For obtaining the probability of choosing the best solution using the roulette wheel; /> Best vulture for the current African vulture;

Phase Two: Calculating Vulture Hunger Rate

Among them, F is the vulture hunger rate, Indicates the current iteration number, /> Indicates the maximum number of iterations, zz is a random number between -1 and 1 and changes every iteration, c is a random number between -2 and 2, /> It is a random number between 0 and 1. When zz falls below 0, it means that the vulture is hungry. If the value of zz increases to 0, it means that the vulture is full. When /> When it is greater than 1, vultures are looking for food in different areas, and the African vulture algorithm enters the exploration stage; when /> When it is less than 1, the African vulture algorithm enters the development stage, and the vulture looks for food near the optimal solution;

Phase Three: Exploration

in, is a random number between -1 and 1, /> is the preset exploration parameter, used to control the exploration strategy, /> The vulture position vector in the current iteration, /> is the vulture position vector in the next iteration, /> and Both are random numbers between -1 and 1, ub and lb are the upper and lower boundaries of optimization;

Phase Four: Development

when Between 0.5 and 1, the African vulture algorithm enters the first sub-phase of the development phase. In the first sub-phase, two different strategies of rotating flight and siege are executed. The strategy is based on the exploration parameters of the first sub-phase /> Make a selection:

in, , /> , /> Both are random numbers between -1 and 1;/> For the first equation of vulture rotational flight, /> The second equation for the rotary flight of the vulture; /> is the distance between the vulture and the best two groups of vultures;

if If it is less than 0.5, execute the second sub-phase of the development phase;

In addition, the development process of the second sub-phase of the development phase is improved by using the Levy flight strategy, the expression is as follows:

Among them, Levy is a flight strategy, xx is a random step size, is an exponential parameter, and u means obeying a normal distribution;

The policy is based on the exploration parameters of the second subphase To select, the improved formula is:

in, is the best vulture of the first set in the current iteration, /> is the best vulture of the second set in the current iteration, /> The first equations of motion for vultures competing for food, /> A second equation of motion for vultures competing for food;/> is a random number between -1 and 1, /> is the flight strategy for the flight distance d ;

The fitness function is:

Among them, hc _is the actual variance of state information; is the innovation sequence of the maximum correlation entropy square root volumetric Kalman filter; /> is the weight coefficient;

After the above-mentioned improved African vulture algorithm is obtained, the non-Gaussian noise of the maximum correlation entropy square root volumetric Kalman filter constructed based on step 3 is optimized, and the maximum correlation entropy square root volumetric Kalman filter is optimized through the improved African vulture algorithm. In this way, the process noise and measurement noise are optimized, and finally the optimal estimation of the state parameters of the four-wheel distributed electric drive vehicle is realized.

Fig. 2 is a comparison chart of the method proposed by the present invention and the traditional volumetric Kalman algorithm and the real value. The double shifting working condition scene is built by carsim software for testing. The mass of the vehicle is set to 1412kg, and the distance from the front axle to the center of mass is 1.015 m, the distance from the rear axle to the center of mass is 1.895m, the longitudinal vehicle speed is 40km/h, the road surface adhesion coefficient is 0.85, and the wheel radius is 0.325m. It can be seen from Figure 2 that the method proposed by the present invention can estimate the side slip angle of the center of mass. Significantly improved, the accuracy increased by about 10%.

In the description of this specification, descriptions referring to the terms "one embodiment", "some embodiments", "example", "specific examples", or "some examples" mean that specific features described in connection with the embodiment or example , structure, material or characteristic is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

Although the embodiments of the present invention have been shown and described, those skilled in the art can understand that various changes, modifications, substitutions and modifications can be made to these embodiments without departing from the principle and spirit of the present invention. The scope of the invention is defined by the claims and their equivalents.

Claims (6)

1. A four-wheel distributed electric drive vehicle state estimation method, is characterized in that, comprises the following steps: Step 1. According to the motion characteristics of the distributed electric drive vehicle, construct a longitudinal, lateral, and yaw nonlinear three-degree-of-freedom vehicle dynamics model, and construct a state-space equation based on the nonlinear three-degree-of-freedom vehicle model; Step 2, based on the state space equation constructed in step 1, connect the square root volumetric Kalman filter with the vehicle dynamics system, establish the discrete state equation and observation equation of the square root volumetric Kalman filter of the nonlinear system, and determine the square root volumetric Kalman filter The estimated quantity, input quantity and observed quantity of the Mann filter, and the square root volumetric Kalman filter is used to correct and recursively update the estimated parameters of the square root volumetric Kalman filter; Step 3, based on the square root volumetric Kalman filter constructed in step 2, using the maximum correlation entropy criterion as an optimization standard to construct a maximum correlation entropy square root volumetric Kalman filter; Step 4, obtain the improved African vulture algorithm, and optimize the non-Gaussian noise of the maximum correlation entropy square root volumetric Kalman filter based on step 3, and optimize the maximum correlation entropy square root volumetric Kalman filter through the improved African vulture algorithm Finally, the optimal estimation of the state parameters of the four-wheel distributed electric drive vehicle is realized. 2. The four-wheel distributed electric drive vehicle state estimation method according to claim 1, characterized in that, in step 1, the nonlinear three freedoms established according to the longitudinal motion, lateral motion, and yaw motion of the distributed electric drive vehicle The vehicle dynamics equation is shown in the following formula: in, , /> are vehicle yaw rate and side slip angle of center of mass respectively; /> represents the front wheel rotation angle; v _x , a _x , a _y represent the longitudinal velocity, longitudinal acceleration and lateral acceleration of the vehicle respectively; k ₁ , k ₂ represent the equivalent cornering stiffness of the front axle and rear axle of the vehicle respectively; is the distance from the front axle to the center of mass, and the distance from the rear axle to the center of mass; I _z is the moment of inertia of the center of mass; m is the mass of the vehicle; /> expresses the differentiation of the indicated quantity; The state space equation of the established nonlinear three-degree-of-freedom vehicle model is as follows: . 3. The four-wheel distributed electric drive vehicle state estimation method according to claim 2, wherein step 2 specifically comprises: The discrete state equation and observation equation of the square root volumetric Kalman filter to construct the nonlinear system are as follows: in, is the state variable at time k , /> , T represents the transpose operation; /> is the control variable at time k , /> ;/> is the observed variable at time k , /> ;/> is the state variable at time k -1; /> is the control variable at time k -1; f is the evolution equation of the discrete state equation; h is the evolution equation of the observation equation, /> and /> are uncorrelated process noise and measurement noise with zero mean, respectively; The forward Euler discretization method is used to obtain the discretized state equation and measurement equation, as shown in the following formula: in, , /> are respectively the vehicle yaw rate and the side slip angle of the center of mass at time k ; /> , /> Respectively represent the longitudinal velocity and lateral acceleration of the vehicle at time k ; /> , /> are the vehicle yaw rate and the side slip angle of the center of mass at time k -1 respectively; /> , /> , /> Respectively represent the longitudinal velocity, longitudinal acceleration and lateral acceleration of the vehicle at time k -1; /> Indicates the front wheel angle of the vehicle at time k -1; The discretized state equation and measurement equation are introduced into the square root volumetric Kalman filter for prediction and update. 4. The four-wheel distributed electric drive vehicle state estimation method according to claim 3, characterized in that, introducing the discretized state equation and measurement equation into the square root volumetric Kalman filter, and performing prediction and updating specifically include: 1) Forecast: At time k , define is the state covariance matrix at time k -1 /> the square root of is the covariance of the process noise /> square root factor of , /> is the covariance of the measurement noise /> square root factor of Calculation of state volume points: in, is the state volume point, /> is the i-th volume point, /> , n is a natural number greater than or equal to 1, /> is the estimated state value at time k -1; Propagation of state volume points: in, is the predicted value corresponding to the volume point in the prediction process; The prior state and the square root of the covariance matrix are estimated by: in, is the state variable at time k predicted according to time k -1; in, is the square root of the state covariance matrix at k time predicted according to k -1 time, /> represents the triangular decomposition of a matrix, /> is the deviation matrix of the state vector; 2) Update: Calculation of measuring volume points: in, The state vector corresponding to the volume point reconstructed for the update phase, /> is the state vector at time k estimated according to time k -1 in the update phase, /> is the square root of the state covariance matrix at time k -1 of the update reconstruction stage; Measure the spread of volume points: in, is the predicted value corresponding to the volume point during the measurement update process; The a priori measurements and the square root of the covariance matrix are estimated by: in, is the measured variable at time k predicted according to time k -1; in, is the measurement covariance matrix at k time predicted according to k -1 time, /> is the deviation matrix of the measurement vector; Compute the cross-covariance matrix: in, is the transpose matrix of the bias matrix of the measurement vector, /> is the cross-covariance matrix, is the deviation matrix of the state vector; Calculate the Kalman gain : in, is the square root factor of the innovation variance, /> is the transposed matrix of the square root factor of the innovation variance; The posterior state and the square root of the covariance matrix are estimated by: in, is the expected state variable at time k , /> is the measured variable collected at time k ; in, is the square root of the state covariance matrix at time k . 5. The four-wheel distributed electric drive vehicle state estimation method according to claim 4, wherein step 3 specifically comprises: Combining the nonlinear three-degree-of-freedom vehicle dynamics model constructed in step 1 with the square root volumetric Kalman filter constructed in step 2, the maximum correlation entropy criterion is introduced to construct a nonlinear recursive model: in, for the process variable, /> , /> covariance of for: in, is the process variable corresponding to the state error covariance, /> is the process variable corresponding to the measurement error covariance, /> is the state covariance matrix at time k ; Then the best estimate based on the maximum correlation entropy criterion is obtained as follows: in, is the state quantity best estimated by the maximum correlation entropy criterion, /> is the kernel function, q + m is the dimension; Then, define is the process variable of the maximum correlation entropy criterion, and the following formula is obtained: in, for /> Diagonal matrix of the first q elements, /> for /> The diagonal matrix of the q +1 to q + m elements of ; Will The updated covariance matrix of is defined as: in, is the state covariance matrix at time k after the maximum correlation entropy improvement, /> The measurement error covariance matrix improved by the maximum correlation entropy; /> for /> the inverse matrix; The prior measurement noise variance can be expressed as: in, for /> inverse matrix; Then, the square root of the updated measurement covariance matrix Obtained by the following formula: Finally, the into the square root volumetric Kalman filter instead of /> An iterative cycle is performed to construct a maximum correlation entropy square root volumetric Kalman filter. 6. The four-wheel distributed electric drive vehicle state estimation method according to claim 5, characterized in that, in step 4, the African vulture algorithm is improved as follows: The first stage: determine the best vulture in the group The chaotic optimization mode is introduced into the African condor optimization algorithm, and the principle of chaos theory is used to explore the search space. The expression is as follows: in, is the global best position at the current moment, /> The global best position for the next moment; After initializing the population, calculate the population fitness value: in, is the probability that other vultures move to the optimal vulture position, /> and /> is the parameter given before the search operation, with a value between 0 and 1, and /> and /> The sum is 1; /> , /> Two groups of best vultures respectively; For obtaining the probability of choosing the best solution using the roulette wheel; /> Best vulture for the current African vulture; Phase Two: Calculating Vulture Hunger Rate Among them, F is the vulture hunger rate, Indicates the current iteration number, /> Indicates the maximum number of iterations, zz is a random number between -1 and 1 and changes every iteration, c is a random number between -2 and 2, /> It is a random number between 0 and 1. When zz falls below 0, it means that the vulture is hungry. If the value of zz increases to 0, it means that the vulture is full. When /> When it is greater than 1, vultures are looking for food in different areas, and the African vulture algorithm enters the exploration stage; when /> When it is less than 1, the African vulture algorithm enters the development stage, and the vulture looks for food near the optimal solution; Phase 3: Exploration in, is a random number between -1 and 1, /> is the preset exploration parameter, used to control the exploration strategy, /> The vulture position vector in the current iteration, /> is the vulture position vector in the next iteration, /> and /> Both are random numbers between -1 and 1, ub and lb are the upper and lower boundaries of optimization; Phase Four: Development when Between 0.5 and 1, the African vulture algorithm enters the first sub-phase of the development phase. In the first sub-phase, two different strategies of rotating flight and siege are executed. The strategy is based on the exploration parameters of the first sub-phase /> Make a selection: in, , /> , /> Both are random numbers between -1 and 1;/> For the first equation of vulture rotational flight, /> The second equation for the rotary flight of the vulture; /> is the distance between the vulture and the best two groups of vultures; if If it is less than 0.5, execute the second sub-phase of the development phase; In addition, the development process of the second sub-phase of the development phase is improved by using the Levy flight strategy, the expression is as follows: Among them, Levy is a flight strategy, xx is a random step size, is an exponential parameter, and u means obeying a normal distribution; The policy is based on the exploration parameters of the second subphase To select, the improved formula is: in, is the best vulture of the first set in the current iteration, /> is the best vulture of the second set in the current iteration, /> The first equations of motion for vultures competing for food, /> A second equation of motion for vultures competing for food;/> is a random number between -1 and 1, /> is the flight strategy for the flight distance d ; The fitness function is: Among them, hc _is the actual variance of state information; is the innovation sequence of the maximum correlation entropy square root volumetric Kalman filter; /> is the weight coefficient; After the above-mentioned improved African vulture algorithm is obtained, the non-Gaussian noise of the maximum correlation entropy square root volumetric Kalman filter constructed based on step 3 is optimized, and the maximum correlation entropy square root volumetric Kalman filter is optimized through the improved African vulture algorithm. In this way, the process noise and measurement noise are optimized, and finally the optimal estimation of the state parameters of the four-wheel distributed electric drive vehicle is realized.