A SELF-LEARNING AND TUNING FUZZY-LOGIC CONTROLLER BASED ON GENETIC ALGORITHMS AND REINFORCEMENTS

This article presents a new method for learning and tuning a fuzzy log ic controller automatically. A reinforcement learning and a genetic al gorithm are used in conjunction with a multilayer neural network model of a fuzzy logic controller, which can automatically generate the fuz zy control rules and refine the membership functions at the same time to optimize the final system's performance. In particular, the self-le arning and tuning fuzzy logic controller based on genetic algorithms a nd reinforcement learning architecture, which is called a Stretched Ge netic Reinforcement Fuzzy Logic Controller (SGRFLC), proposed here, ca n also learn fuzzy logic control rules even when only weak information , such as a binary target of ''success'' or ''failure'' signal, is ava ilable. We extend the AHC algorithm of Barto, Sutton, and Anderson to include the prior control knowledge of human operators. It is shown th at the system can solve a fairly difficult control learning problem mo re concretely, the task is a cart-pole balancing system, in which a po le is hinged to a movable cart to which a continuously variable contro l force is applied. (C) 1997 John Wiley & Sons, Inc.