Robust self-learning fuzzy controller design for a class of nonlinear MIMOsystems

A new learning paradigm that can be applied for the design of a fuzzy logic -based learning controller that is robust to external signals such as distu rbances and set-point changes is proposed in the paper. It is well known th at the self-organizing fuzzy controller proposed by Procyk and Mamdani is s ensitive to external signals. Such a phenomenon may be observed in other ty pes of direct fuzzy logic-based learning controllers that utilize an adapta tion scheme in which the locational information of the current error state vector determines the degree of adaptation that should be made in the learn ing controller. To resolve the problem of sensitivity to external signals, it is proposed that learning and modification of the controller be made in consideration of the motional trend of the error state vector as well as it s locational information. This paradigm is adopted in the design of a new r obust self-learning fuzzy controller for a dass of nonlinear MIMO systems. The well-known techniques of sliding mode control and the fuzzy decision ma king method are utilized to implement the proposed learning scheme in the f uzzy learning controller. Via simulation study and experimental results, th e proposed learning controller is verified to be robust in the presence of external disturbances. (C) 2000 Elsevier Science B.V. All rights reserved.