Title

Optimal Sampling and Regulation of Uncertain Interconnected Linear Continuous Time Systems

Abstract

This paper presents a co-design approach for optimizing both the sampling period and the control policy of uncertain isolated and interconnected linear continuous time systems. The simultaneous optimization of event-based aperiodic sampling instants and control policy is formulated as a min-max problem and a saddle point solution is generated. An adaptive solution to the min-max control problem is obtained using a two player zero-sum game based Q-leaning approach. Novel update laws, both for flow period and jump instants, for updating the Q-function parameters are proposed in an impulsive system framework. Asymptotic regulation of the system states and Q-function parameters are guaranteed for both the isolated and interconnected systems, under the assumption of persistence of excitation (PE) condition. It is demonstrated that the resulting sampled data implementation of the controllers for both the isolated and interconnected systems do not exhibit Zeno behavior. Numerical results are included to substantiate the claims.

Meeting Name

2017 IEEE Symposium Series on Computational Intelligence, SSCI (2017: Nov. 27-Dec. 1, Honolulu, HI)

Department(s)

Electrical and Computer Engineering

Comments

This research is supported by NSF grant 1406533 and Intelligent Systems Center, Rolla and Startup fund Oklahoma State University, Stillwater, OK.

Keywords and Phrases

Artificial intelligence; Impulse response; Lyapunov methods; Adaptive solution; Asymptotic regulation; Co-design approach; Impulsive systems; Linear continuous-time system; Numerical results; Persistence of excitation; Simultaneous optimization; Continuous time systems

International Standard Book Number (ISBN)

978-1-5386-2726-6

Document Type

Article - Conference proceedings

Document Version

Citation

File Type

text

Language(s)

English

Rights

© 2017 Institute of Electrical and Electronics Engineers (IEEE), All rights reserved.

Share

 
COinS