A Design on Adaptive Two-Step Filter Applied in Nonlinear System

Abstract

Two-step filter is a filter that defines a set of new state parameters. These state parameters are nonlinear functions reflecting real-time states of system. Then, applying Kalman filter to acquire best estimate value, we can calculate best answer by Gauss-Newton alternate arithmetic. Not knowing real noise statistic values, we should combine variable noise statistic estimator with two-step filter. Applying this method is called adaptive nonlinear filter. Analyzing simulation results of adaptive two-step filter, it is concluded that performance of adaptive two-step filter exceeds one of other filters.