A Novel Mixture Least Squares Approach for Simultaneous Parameter/State and Unknown Input Estimation

• 成果类型: Article
• 署名作者: Ding, Bo; Wei, Yuanchu; Zhang, Yong; Yang, Wu
• 署名单位: Yangzhou University; Wuhan University of Science & Technology; Guangxi University
• 刊物名称: IEEE TRANSACTIONS ON AUTOMATIC CONTROL
• ISSN/ISSBN: 0018-9286
• DOI: 10.1109/TAC.2024.3450001
• 发表日期: 2025
• 页码: 1252-1258
• 关键词: Covariance matrices; estimation; Stochastic processes; Maximum likelihood estimation; Kalman filters; vectors; Filtering algorithms; Linear stochastic systems; mixture least squares (MLSs); parameter/state estimation; unknown input estimation

摘要

In this article, we introduce a novel mixture least squares (MLSs) algorithm to deal with the problems of simultaneous parameter/state and unknown input estimation. First, the MLSs algorithm is derived to estimate the desired parameter and unknown input, which can be regarded as a unified framework for deterministic least squares and stochastic least squares. The unbiasedness and optimality of the MLSs estimators are further verified. Then, based on the established MLSs algorithm, a new solution to simultaneous state and unknown input estimation (SUIE) problems is given. The proposed method is more concise and straightforward than the existing SUIE algorithms. The method provided in this article offers fresh insight into parameter/state estimation with unknown input.