Completely Distributed Joint State-Unknown Input Estimation via Interval Observer

• 成果类型: Article
• 署名作者: Luo, Miaohong; Su, Housheng; Zeng, Zhigang
• 署名单位: Huazhong University of Science & Technology; Zhengzhou University
• 刊物名称: IEEE TRANSACTIONS ON AUTOMATIC CONTROL
• ISSN/ISSBN: 0018-9286
• DOI: 10.1109/TAC.2026.3662304
• 发表日期: 2026
• 关键词: FAULT-DETECTION; DESIGN; ORDER

摘要

This article investigates the design of a novel completely distributed unknown input observer (DUIO) for linear time-invariant systems without relying on any global information. Existing methods face significant limitations, such as difficulty in simultaneously achieving asymptotic estimation of both system state and unknown input, substantial increases in observer orders, or even the nonexistence of feasible observers. To address these challenges, a novel DUIO design framework based on interval observers is developed. Specifically, the proposed method begins by employing a detectability decomposition technique to divide a target system into detectable and undetectable subsystems. Next, a time-varying coordinate transformation is designed for the detectable subsystem to construct interval observers, from which algebraic unknown input functions are derived. Subsequently, dynamic coupling gains are introduced to develop a completely DUIO, consisting of an adaptive distributed observer and an unknown input reconstruction. Finally, the effectiveness of the proposed approach is validated through practical applications, demonstrating its potential value in real-world engineering scenarios.