Parameter Identification in Switching Multiple Model Estimation and Adaptive Interacting Multiple Model Estimator
具有参数自适应的交互式多模型算法

Switching multiple model estimation (SMME) has been widely applied in problems with both structural and parametric uncertainties and/or changes, ranging from target tracking to fault detection and isolation. However its filtering parameters, determined by a priori information, are the tradeoff between the "mode transition" case and the "non mode transition" case. Hence an online method for SMME to identify filtering parameters, including Markov transition probabilities and the variances of model conditional process noise, are proposed, by using additional multiple state predictors at the beginning of each filtering cycle. By combining the parameter identification with interacting multiple model (IMM), which is one of the most cost effective estimators in SMME, we present an adaptive IMM (AIMM), which shows much more accurate than IMM in the simulation of tracking a maneuvering target.