Research And Application Of Constrained State Estimation Based On UHF

The traditional algorithms cannot handle the constrained state estimation problem of nonlinear non-Gaussian systems well.Therefore,the problems of state estimation,state constraints and parameter optimization of nonlinear non-Gaussian systems are deeply explored in this paper,and new solutions based on UHF algorithm are proposed.The main contents of this paper are as follows:Firstly,compared with traditional state estimation algorithms,UHF algorithm can realize the better state estimation of nonlinear non-Gaussian systems,and effectively resist the influence of noise on state estimation.The paper verifies this conclusion at first.Secondly,this paper discusses the traditional types of state constraints in the state estimation of nonlinear systems and their processing methods.Then the projection method and the boundary shrinkage method are proposed for the UHF algorithm.The projection method always projects the sigma points that violate the state constraints onto the constraint boundary.In the boundary shrinkage method,the sigma points will be scaled together to fit the constraint surface if there exist some points that violate the state constraints.Both of these processing methods complete the preliminary preparation of the UHF algorithm in the restricted state estimation of the nonlinear system.The experiments show that the UHF algorithm with constraints can achieve better results than that of the traditional UHF algorithm.Thirdly,the paper further studies the influence of the sigma point set distribution in the UHF algorithm on the performance of state estimation,and introduces the particle swarm optimization algorithm(Particle Swarm Optimization,PSO)to realize the adaptive learning of the sigma point set generation parameters,so as to obtain the best sigma point set.Experimental results show that this method can improve the state estimation accuracy of UHF algorithm.In this paper,the hardware platform of State Of Charge(SOC)of lithium battery for electric vehicle is built by ourselves,and various original data sets of batteries are tested and extracted.The application of improved UHF algorithm in SOC estimation is tested and evaluated by using this data set and lithium iron phosphate battery data set provided by Hangzhou Wanxiang Group.Experimental results show that the mean square error of SOC estimation based on the improved UHF algorithm can reach 9.1608e-6,which is two orders of magnitude higher than the conventional algorithm,thus verifying the effectiveness of the optimization algorithm.