Variety of meta-heuristics based on genetic algorithms to solve a generalized job-shop problem

In this paper we address a generalized job-shop scheduling problem with unrelated parallel machines and precedence constraints between the jobs operations (corresponding to either linear or non linear process routings). The objective is to minimize the completion time. Resolution of several scheduling problems, including parallel machines scheduling is referred as NP-hard. So, the application of approximate methods to solve them is well appropriate. Considering the success of genetic algorithms, we develop a variety of original techniques based on this meta-heuristic to solve the considered problem. These techniques integrate different strategies linked to mutations and crossovers for selecting the individuals for reproduction and generating a new population. The performance of these algorithms is tested by numerical experiments using randomly generated benchmarks. A comparison between the considered meta-heuristics results is presented.