Fault Detection for Stochastic Switched System Based on Fuzzy Adaptive Unscented Kalman Filter

摘要

In order to solve the problem of state estimation and fault detection for a class of stochastic switched nonlinear system with unknown characteristics of observation noise change, a system state estimation algorithm based on interactive multiple model and unscented Kalman filter (IMM-UKF) is proposed. The algorithm uses the UKF to estimate the state of each subsystem at different time points, then fuses the state estimation results of different subsystems to obtain the final state estimation, and realizes the estimation of the real state of the system. In order to achieve real-time tracking of system observation noise, a fuzzy controller is established to adjust the observation noise of IMM- UKF in real time, and a fuzzy adaptive IMM-UKF algorithm (FAIMM-UKF) is proposed. For actuator failure in a class of stochastic switched nonlinear systems, FAIMM-UKF is used to estimate the system state. Based on the result of the state estimation, residual and residual evaluation functions are established to detect the actuator fault. Finally, the effectiveness of the proposed algorithm is verified by simulation experiment.