Parameter identification of nonlinear system and its application based on strong tracking filter and wavelet transform

The strong tracking extended Kalman filter (STEKF) is used as the main frame and the linearization and state expansion are employed to estimate the time-varying parameters and states of nonlinear systems. Based on the general STEKF, a wavelet-transform-based filter is proposed to estimate the variance of the measurement noise, and a new filtering gain factor is utilized in STEKF to eliminate the tracking overshoot. Main formulas for calculation and the methods for selecting parameters are presented. Monte Carlo simulation and practical application in identification of ballistic parameters demonstrate that the proposed method can exactly estimate the abruptly changing parameters even when the variance of the measurement noise is time-varying. The estimation accuracy of parameters and states is higher than that of the general STEKF.