On-Line Search for Optimal Controller Parameters Using Interconnected LearningAutomata

摘要

The application of interconnected learning automata to control of unknownstochastic dynamic systems is addressed. A method of using N learning automata and subsets of actions to search controller parameters according to a given performance index is presented. Simulation results of evaluation on this method for control of an unknown turbogenerator system are included. The results demonstrated potential for control of unknown stochastic dynamic systems.