A steady-state evolutionary algorithm for the job shop problem

This paper describes an evolutionary search scheduling algorithm (ESSA) for the job shop scheduling problem (JSSP). If no specific knowledge of the problem is included in the EA then it is less likely that an EA will find the global optimum of a JSSP. However, the proposed steady state ESSA can find global optima if a particular population and offspring size is utilized. Adding specific knowledge through a Lamarckian Learning process improves the performance significantly. In this case, a good balance of the amount of Lamarckian Learning can realize a good performance in both efficiency and effectiveness.