Fuzzy genetic algorithms based on fuzzy number coding on [0, 1] and its application

Abstract

An improved genetic algorithm for fuzzy programming problems with triangular fuzzy variables is proposed in this paper. The decentralization degree of fuzzy numbers is defined, and the way of coding triangular fuzzy number on [0, 1] is built. So we can limit the type of fuzzy numbers, increase the convergence property of algorithm, and make the fuzzy information processing more reasonable. Based on the structure characteristics of optimization variables, the crossover operation was replaced by linear recombination, and a compound mutation operation to triangular fuzzy numbers is given. The effectiveness and usefulness are discussed through an example in the end.