Fusion and hierarchy can help fuzzy logic controller designers

In this paper, in the context of fuzzy control, the problem of rule-base explosion is addressed. Indeed if we want to control a multivariable system it is urgent to reduce the fuzzy logic controller (FLC) complexity in order to facilitate its synthesis. First some rule-base reduction methods like fusion and hierarchy are applied in order to define a simpler FLC structure. The FLC is then composed of sub FLCs having at the very most, just 2 inputs and 1 output The number of parameters is then considerably reduced. Then, optimization techniques like the gradient descent method can be used to tune them. Validation has been done on a two-input-two-output mixing tank