State Parameters Estimation Of Distributed Electric Vehicle Based On Multi-Sensor Fusion

In recent years,as one of the important means to realize energy transformation in China,the electric vehicle industry has developed vigorously.The distributed drive electric vehicle is an important type of electric vehicle.Compared with traditional centralized drive electric vehicle,its motor driving/braking torque is independently controllable,with high control precision and faster response.The distributed electric drive is the main trend in the development of smart car power transmission,and the accurate acquisition of vehicle driving state parameters is the basis for correct decision-making and control of smart cars.Due to technical and cost problems,it is difficult to obtain real-time data of some vehicle state parameters such as the side slip angle directly through sensors.Therefore,it is necessary to study an algorithm with high accuracy and strong stability to estimate it in real time.Accurately estimating the driving state parameters of distributed electric vehicles has always been a research hotspot of scholars at home and abroad.At present,the main algorithms used to estimate vehicle state parameters are the nonlinear observer,the synovium observer and the Kalman filter observer,among which the most widely used algorithm in the vehicle state parameter estimation is Kalman filter algorithm and its improved algorithm.Aiming at the problem of parameter estimation of the driving state of distributed electric vehicles,the main research work of this paper is as follows:(1)The research background of distributed drive electric vehicle state parameter estimation is introduced in detail,and the high-order cubature Kalman filter is used as the research basis of the estimation algorithm.Secondly,the nonlinear dynamics formula of the vehicle is analyzed,a distributed drive electric vehicle co-simulation platform is built on Simulink/Carsim software.(2)The traditional CKF is improved by using the generalized cubature integral instead of the sphere-phase integral to directly obtain the cubature points and weights of the algorithm,and improved the high-order scalability of the CKF and the robustness to multi-dimensional systems.(3)Use the orthogonal trigonometric decomposition instead of the Cholesky decomposition.The Cholesky decomposition is used to solve the error covariance matrix P_k,and it is easy to generate an indefinite matrix to stop the Cholesky decomposition and cause the algorithm to collapse.In order to improve the overall stability of the algorithm,this paper introduces the idea of the square root filtering,and replaces the Cholesky decomposition with the QR decomposition,which improves the estimation accuracy while ensuring the stability of data iteration.(4)For the local fusion problem of the multi-sensor system of the vehicle,the cross-covariance matrix is often difficult to obtain.Therefore,this paper uses the ICI fusion algorithm to reduce the problem of repeated calculation of the unknown public information and improve the estimation accuracy of the algorithm when the cross-covariance between sensors is unknown.(5)During the normal driving of the vehicle,both the process noise covariance matrix Q and the measurement noise covariance matrix R of the system are unknown noises,and the Q matrix and R matrix play a crucial role in the estimation accuracy of the algorithm.Therefore,this study uses the atomic search algorithm(ASO)to seek the optimal noise value for the noise covariance matrix(Q,R)to improve the estimation accuracy of the algorithm.(6)On the simulation and real vehicle experimental platform,the double line shifting condition experiment and the serpentine condition experiment were carried out,and the data collected by the sensors during the experiment were compared with the estimated value of the algorithm,which further verifies the effectiveness and robustness of the state parameter estimation algorithm proposed in this paper.