Identification And State Estimation Of Polynomial Systems Using Extended Kalman Filtering And Fuzzy Multiobjective Evolutionary Algorithms

In recent years, identification of polynomial system is a hot research topic in the field of nonlinear system identification.Aiming to solve the problem of the accuracy of the states and parameters estimation are greatly influenced by initial values in the polynomial systems, this paper proposes a extended kalman filtering (EKF) based joint state estimation and parameter identification method in the polynomial systems. First, using the least squares method to estimate the parameters of the polynomial system. Let the results of the least square as the initial values in EKF to estimate the states and parameters jointly in the polynomial systems. The results show that compared with the results obtained by using EKF, results obtained by the proposed method can greatly reduce the system state estimation error covariance.At the same time, this paper proposes a novel method based fuzzy multiobjective evolutionary algorithm for the identification of the polynomial systems. First, choose the first order monomials by using the correlation-based orthogonal forward search algorithm. Then according to the correlation coefficient of monomial term and output and the contribution of each monomial term to system output to narrow the scope of the candidate monomials. At last, change the pareto dominance into fuzzy one, accordingly change the fast-non-dominated-sort algorithm into fuzzy-fast-non-donminated-sort algorithm. Apply the fuzzy multi-objective evolutionary algorithm to get the identification results. The results show that compared with the nondominated sorting genetic algorithm II and Akaike information criterion, the proposed method can get only one solution and at the same time it can get higher accuracy for identification of the system structure.