Neuro-Fuzzy Approaches for Identification and Control of Nonlinear Systems

Neural Networks and Fuzzy Inference Systems are becoming well-recognized tools of designing an identifier/controller capable of perceiving the operating environment and imitating a human operator with high performance. The motivation behind the use of neuro-fuzzy approaches is based on the complexity of real life systems, ambiguities on sensory information or timevarying nature of the system under investigation. In this respect, neuro-fuzzy design approaches combine architectural (by neural networks) and philosophical (by fuzzy systems) aspects of an expert resulting in an artificial brain, which can be used as an identifier or a controller. It is known that the fuzzy inference systems and neural networks are universal approximators. An architecture with an appropriate learning strategy can teach any mapping to such a system with a predefined realization error bound. The most questionable quality in the use of neurofuzzy architectures is the stable training. This tutorial considers various neuro-fuzzy structures and gradient based training procedures. Consideration is given to stabilization of training dynamics.