A New Extended Set-Membership State Estimation Algorithm And Application

Set-Membership Estimation theory has become a new and highly active research in the fields of International Automatic Control theory, which has already got the extensive application in system identification, information processing, fault detection and robust control. Based on the existing achievements of Set-Membership algorithm and its application, further research work is carried out in this paper, which is mainly focus on analyzing the state estimation for nonlinear discrete-time system.The paper first introduced several common state estimation algorithms of nonlinear system. Weighted Least Square, Extended Kalman Filter and Extended Set-Membership Estimation which is based on the outer bounding ellipsoid, have been introduced respectively. And then discusses the advantage and disadvantage of each algorithm and compares the difference among them.Based on the existing achievements on Set-Membership algorithm, a New Extended Set-Membership Estimation algorithm for nonlinear discrete-time system is presented in this paper. This algorithm is simple in description, good in real time capability, and robust in disturbance of parameter. The proposed algorithm, which is based on state bounding techniques that is guaranteed to contain the true state of the system, can be decomposed into two steps: time updating and observation updating. In addition, the algorithm is not only linearized the equations of the nonlinear system, but also get the first and second order of the linearization errors together to analysis. It is proved that under certain assumptions the estimation error remain bounded. Simulation results have shown the efficiency and robustness of the proposed algorithm.Finally, the application of the New Extended Set-Membership algorithm in Power System State Estimation is introduced. The characteristic and performance of some common algorithms for Power System State Estimation are compared with the New Extended Set-Membership algorithm through the simulation. The experiment results showed that the New Extended Set-Membership algorithm is effective, and it can raise the accuracy of the state estimation.