A multi-operator self-tuning genetic algorithm for fuzzy control rule optimization

We propose a multi-operator self-tuning GA (MSGA) for fuzzy controller optimization problem. The MSGA is designed with multiple operators-single-point crossover (CR), single-point mutation (MU), single-point copy (CO) and single-point exchange (EX). The MU, CO, and EX operators are considered as a group of mutation because they select one parent and generate one offspring at one time of reproduction. Therefore, we have two groups of operators, one is simple crossover only and the other is a group of mutation operators. These two groups of operators will do the search job repeatedly until it reaches predetermined stop conditions. In the mutation loop, a fuzzy reasoning mechanism is applied to decide the population size of each mutation operator by sensing searching capabilities of each operator in different stage.