Bottleneck identification procedures for the job shop scheduling problem with applications to genetic algorithms

Abstract

Two bottleneck identification algorithms (one for bottleneck machines and the other for bottleneck jobs) are presented for the job shop scheduling problem in which the total weighted tardiness must be minimized. The scheduling policies on bottleneck machines can have significant impact on the final scheduling performance, and therefore, they need to be optimized with more computational effort. Meanwhile, bottleneck jobs that can cause considerable deterioration to the solution quality also need to be considered with higher priority. In order to describe the characteristic information concerning such bottleneck machines and bottleneck jobs, a statistical approach is devised to obtain the bottleneck characteristic values for each machine, and, in addition, a fuzzy inference system is employed to transform human knowledge into the bottleneck characteristic values for each job. These bottleneck characteristic values reflect the features of both the objective function and the current optimization stage. Finally, the effectiveness of the two procedures is verified by specifically designed genetic algorithms.