Adaptive Control Using Collective Information Obtained from Multiple Models*

A radically new approach to adaptive control using multiple models was recently proposed by the authors. Using parametric estimates and outputs generated by multiple adaptive identification models, an adaptive control procedure was proposed which resulted in significantly faster and more accurate response of the overall system, as compared to existing methods such as “switching” and “switching and tuning”, which also use multiple models. In this paper the ideas are extended to include fixed identification models. Further, many of the theoretical questions that have arisen concerning the stability of the overall system and the reasons for the improved performance are addressed in detail. Simulation results are included, towards the end of the paper, to indicate the effectiveness of the new methodology.