Self-Tuning Methods for Multiple-Controller Systems

The optimization of stochastic systems with unknown parameters and multiple decision-makers or controllers each having his own objective is considered. Based on a centralized information pattern, steady-state solutions are obtained for the stochastic adaptive Nash game and Leader-Follower game problems. These adaptive solutions, after a judicious transformation, resemble closely the implicit self-tuning solution for the single-controller single-objective case, and thus preserve the salient and advantageous features of self-tuning methods--simplicity and easy implementation. In addition, due to this close resemblance, convergence results for the game problems are established by extending the convergence result from the single-controller single-objective case. The decentralized stochastic adaptive Nash game problem is also considered. Two explicit self-tuning type algorithms are proposed. The first algorithm is an ad hoc constraint on the policy form while the second one is based on extension from static Nash game theory. Simulation results indicate all these self-tuning methods are capable of stabilizing a system along targeted paths.

Name thesis:degree_name

Ph.D.

Level thesis:degree_level

Dissertation

Discipline thesis:degree_discipline

Electrical Engineering

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2014

Author dc:creator

• Chan, Yick Man

• Engineering, Electronics and Electrical

Language dc:language

eng

Identifier

(UMI)AAI8203422

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/66258