An improved self-structuring neuro-fuzzy algorithm
Jie Su, Jia Ren, Haipeng Pan · 2008
Self-organizing fuzzy control (SOFC) is becoming an attractive technique to control processes due to its ability to solve problems in the absence of an accurate mathematical model and the independence of experts. This paper presents a novel approach, which combines both self-structuring fuzzy control algorithm and neural network (NN). The self-structuring fuzzy control can adjust the number of membership functions and rules when needed. The initial fuzzy system is of a small number of membership functions and rules. To confine the size of fuzzy system from growing indefinitely, the algorithm replaces old membership functions by new ones instead of adding more membership functions so that the number of rules never exceeds a predefined upper bound. Then, the radial basis function (RBF) neural network is used to optimize the parameters of self-structuring fuzzy control. Simulations confirm that the characteristics of the new proposed algorithm have improved notably compared with fuzzy control and neuro-fuzzy control.