A hybrid intelligent controller based on 'consultation of doctors'

摘要

How to effectively integrate different intelligent controllers has become an interesting issue in the field of intelligent control. In this paper, a framework of a hybrid intelligent control system is presented, in which multiple different controllers are effectively integrated together and to gain better control quality, working on the principle of "the consultation of doctors". An on-line approach analogous to human debating and compromising is adopted to coordinate different controllers to overcome their respective shortcomings by learning the other's strong points via a coordinator. Performance evaluation is made based on the hybrid controller consisting of an expert controller and a neurocontroller. Knowledge transference and sharing between the expert controller and the neurocontroller is achieved by making use of a mapping between the usable-characteristic vector space and the output vector space. Simulation shows that this approach provides good control quality and high reliability.