Stochastic Filtering Scheme of Implicit Forms of Uncertain Max-Plus Linear Systems

• 成果类型: Article
• 署名作者: Espindola-Winck, Guilherme; Hardouin, Laurent; Lhommeau, Mehdi; Santos-Mendes, Rafael
• 署名单位: Universite d'Angers; Universidade Estadual de Campinas
• 刊物名称: IEEE TRANSACTIONS ON AUTOMATIC CONTROL
• ISSN/ISSBN: 0018-9286
• DOI: 10.1109/TAC.2022.3176841
• 发表日期: 2022
• 页码: 4370-4376
• 关键词: mathematical models; Multiprotocol label switching; Filtering; Linear systems; Algebra; Task analysis; stars; discrete event systems; interval contractors; max-plus linear (MPL) systems; stochastic state estimation; timed event graphs (TEG)

摘要

This article aims to improve the stochastic filtering algorithm with bounded disturbances, proposed in 1. This filter is efficient for max-plus linear systems in explicit form, i.e., the timed event graph (TEG) described by this system is initially with one token on each place. Nevertheless, it needs strong assumptions in order to be accurate for systems in implicit form, i.e., the corresponding TEG is initially with some token-free places which implies that some entries of the system state vector are dependent on each other. In this article, we consider a more general method without these assumptions. It is based on an iterative procedure that widely increases the accuracy of the estimation.