Flexible job-shop scheduling problem by genetic algorithm and learning by partial injection of sequences

The flexible job-shop scheduling problem (FJSP) is one of the most complex problems (Meriem and Ghédira, 2004). Thus, exact methods are not effective for solving the FJSP and heuristic approaches are generally used to find near optimal solutions within reasonable computation times. In our previous works (Fekih and Jallouli, 2010), we have proposed a new approach based on genetic algorithms and the learning by injection of sequences for solving the flexible job-shop scheduling problem (FJSP) with partial or total flexibility. This approach was based on a joint resolution of the inherent assignment subproblem and the sequencing subproblem. In this paper, we develop a new structure of genetic algorithm and defined a new strategy of learning (partial injection of sequences) in order to improve the total duration of the schedule or the makespan. The proposed genetic algorithm uses a new crossover and mutation operators. We called this approach genetic algorithm with learning by partial injection of sequences (GALIS.II). Numerical experiments show that our GALIS.II algorithm is effective for solving the FJSP.