Real-Coded Parameter-Free Genetic Algorithm for Job-Shop Scheduling Problems

摘要

We propose a new genetic algorithm (GA) for job-shop scheduling problems (JSSP) based on the parameter-free GA (PfGA) and parallel distributed PfGA proposed by Sawai et al. The PfGA is not only simple and robust, but also does not need to set almost any genetic parameters in advance that need to be set in other GAs. The performance of PfGA is high for functional optimization problems of 5- or 10-dimensions, but its performance for combinatorial optimization problems, which search space is larger than the functional optimization, has not been investigated. We propose a new algorithm for JSSP based on an extended PfGA, extended to real-coded version. The GA uses random keys for representing permutation of jobs. Simulation results show that the proposed GA can attain high quality solutions for typical benchmark problems without parameter tuning.