A Crossover Operator of Genetic Algorithms for Periodic Function Optimization

Abstract

In engineering, we often face to solve optimization problems of periodic functions having angles as decision variables. For example, a timing optimization for apparatus with cycling mechanism such as internal-combustion engine can be defined as an optimization problem of periodic function. In this paper, first, we discuss problems in optimization of periodic function by genetic algorithms (GAs). Second, to cope with such problems, we propose a new crossover operator for real-coded GAs for periodic function optimization. Finally, we show that the optimization problems of periodic functions are effectively optimized with the proposed method, and it also has some effectiveness in optimization with in bounded region.