Robust Identification of Time-Varying Stochastic Systems

Abstract

This is a second part for robust parameters estimation. Here we shell consider the case of time-varying parameters. First is considered parameters as deterministic which are modeled as random walk. As an estimator the robust Kalman filter is used. As an input signal is considered 1/f signal with corresponding autocovariance function. This signal is suitable for system identification, especially for the case of robust experiment design. In this part the some modifications for robust Kalman filter, as in part 1, are used. The simulations show good behavior of robust real-time identification algorithms.