有观测噪声的时变系统的参数估计

本文给出了有观测噪声、线性离散时变系统的参数估计新方法。它由两段互耦的自适应状态估计器和自适应参数估计器组成。通过引入虚拟时变噪声,我们结合在互耦算法中产生的模型误差到虚拟噪声统计,使模型误差得到有效地补偿和克服滤波发散。模拟例子说明了本文方法的有效性。

Parameter Estimation of Time-varying Systems with Observation Noise

This paper presents a new approach to estimate parameter of time-varying systems with observation noise1]. It consists of two-stage coupled adaptive estimators. The first stage is state estimator by an adaptive recursive filter for systems with multiple time delays2]; The second stage is parameter estimator by an adaptive Kalman filter3]. Introducing the fictitious noise, so that model errors are compensated and effectively overcome filtering divergence. The simulated example is given to show the usefulness of the newly proposed approach.