Linear quadratic Gaussian control of discrete-time Markov jump linear systems with horizon defined by stopping times

• Autor(a) principal: Nespoli, Cristiane
• Data de Publicação: 2005
• Outros Autores: Zúñiga, Yusef R. C., Do Val, João Bosco R.
• Tipo de documento: Artigo de conferência
• Idioma: eng
• Título da fonte: Repositório Institucional da UNESP
• Texto Completo: http://dx.doi.org/10.3182/20050703-6-CZ-1902.00356; http://hdl.handle.net/11449/68594
• topic: Markov models; State and output feedback control; Stopping times; Algebraic Riccati equations; Constructive methods; Control gains; Coupled set; Discrete-time; Dynamic output feedback; Failure events; Linear quadratic Gaussian control; Markov jump linear systems; Markov model; Output feedback controls; Recursions; Separation principle; Stopping time; Algebra; Automation; Control; Linear systems; Markov processes; Riccati equations; Robustness (control systems); State feedback; Discrete time control systems

The linear quadratic Gaussian control of discrete-time Markov jump linear systems is addressed in this paper, first for state feedback, and also for dynamic output feedback using state estimation. in the model studied, the problem horizon is defined by a stopping time τ which represents either, the occurrence of a fix number N of failures or repairs (T N), or the occurrence of a crucial failure event (τ δ), after which the system paralyzed. From the constructive method used here a separation principle holds, and the solutions are given in terms of a Kalman filter and a state feedback sequence of controls. The control gains are obtained by recursions from a set of algebraic Riccati equations for the former case or by a coupled set of algebraic Riccati equation for the latter case. Copyright © 2005 IFAC.