An evolutionary approach to fuzzy rule-based model synthesis using indices for rules (Q1398937)

The author describes a relatively effective genetic algorithm to explicitly build a rule-based fuzzy controller. The algorithm may start from expert knowledge as well as from analytic descriptions of the control process. It generates a list of linguistic control rules assuming that the linguistic input and output variables as well as the linguistic labels for their fuzzy values and some parametric form of their membership functions are given in advance. The algorithm is illustrated with real world examples. The nice point is that the rule base created by the algorithm can really be printed out.