The Perfect Recursive Least Square Algorithm and Applications to Tracking and Prediction

摘要

Based on the theory of generalized inverse, the perfect recursive least square (PRLS) algorithms for the cases of general weighted, optimal weighted and exponetial weighted are derived in this paper. The PRLS algorithms give the least square estimal (LSE) without the requrirnent of a priori statistical knowledge of the initial state to be estimated. Further, with the FRLS algorithms, we provide the deadbeat and unbiased state estimators for some linear systems, and present the multimodel tracking and prediction (MMTP) algorithms for maneuvering