The job-shop scheduling has been a problem for production management ever since the manufacturing industry began. Most scheduling problems are complex combinatorial optimization problems and solve difficultly. This paper presents a new learning algorithm (LA) based on different learning mode for job-shop scheduling problem that the objective is to minimize the makespan. The LA mechanism based on students’ knowledge update the learning from the experience of all seniors and the best of all students collectively. Computational experiments are carried out in which some benchmark instances have been solved to examine the performance of LA for job-shop scheduling problem (JSSP). Experimental results show that LA always outperforms genetic algorithm (GA) remarkably. The LA can be used as a new efficient optimization tool to solve such as TSP, scheduling etc, combinatorial optimization problem.

new learning algorithm, genetic algorithm, PSO, JSSP, scheduling.