A Nonlinear Adaptive Controller: A Comparison between Fuzzy Logic Control and Neurocontrol

This paper explores, from a surface-fitting viewpoint, two algorithmswhich are often applied in the field intelligent control: fuzzyself-organizing controllers and neural networks. Both methodologiesadapt internal model parameters in response to the plant's input-outputmapping. However, while the convergence of single-layer neuralnetworks has been studied in great detail, very few theoremshave been proved about self-organizing fuzzy logic controllers.In this paper, it is shown that B-splines can provide a frameworkfor choosing the shape of the fuzzy sets. Then the operatorschosen to implement the underlying fuzzy logic are examined,showing how they can produce ‘smooth’ control surfaces.It is now possible to make a direct comparison between fuzzylogic controllers and feedforward neural networks, demonstratingthat, in a forward-chaining mode, storing the plant's behaviourin terms of weights or rule confidences is equivalent. Finally,three training rules for the self-organizing fuzzy controllerare derived.