Optimization of Fuzzy Reasoning by Genetic Algorithm Using Variable Bit-Selection Probability

摘要

Genetic algorithms (GA) are known as optimization algorithms that can avoid convergence to local solutions by global search in solution space. However, especially in the field of control, when fuzzy reasoning is to be optimized, global optimal solutions are not necessarily required. In many cases, local solutions obtained at low cost are preferable. For this purpose, GAs are used in combination with other optimization algorithms, for example, steepest descent or pattern search. In so doing, however, there is a problem of differently representing the parameters to be optimized. Besides, complicated software is required to implement such combined methods. A method is proposed in this paper to provide locality in search space by varying the bit-selection probability in GA-based mutations in accord with learning progress. This makes possible local search in the vicinity of good solutions found in the course of optimization, resulting in rapid finding of local optimal solutions.