Approach to Hybrid Flow-Shop Scheduling Problem Based on Self-Guided Genetic Algorithm

Abstract

The effective self-guided genetic algorithm (SGGA) which we proposed is based on the characteristics of a hybrid flow shop scheduling problem. A univariate probability model based on workpiece permutation is introduced together with a bivariate probability model based on a similar workpiece blocks. An approach to updating a probability model parameters is given based on superior individuals. A novel probability calculation function is proposed taking advantages of statistical learning information provided by univariate and bivariate probabilistic model to calculate the probability of workpieces located in different positions. A method for evaluating the quality of individual candidates generated by GA crossover and mutation operators is suggested for selecting promising and excellent individual candidates as offspring. Simulation results show that the SGGA has excellent performance and robustness.