Evolutionary optimization of due date based objectives in unrestricted identical parallel machine scheduling problems

Abstract

Parallel machine scheduling, involves the allocation of jobs to the system resources (a bank of machines in parallel). A basic model consisting of m machines and n jobs is the foundation of more complex models. Here, jobs are allocated according to resource availability following some allocation rule. In the specialised literature, minimisation of the makespan has been extensively approached and benchmarks can be easily found. This is not the case for other important objectives such as the maximum tardiness and the number of tardy jobs. These problems are NP-hard for 2 ≤ m ≤ n, and conventional heuristics and evolutionary algorithms (EAs) have been developed to provide acceptable schedules as solutions. To solve the unrestricted identical parallel machine scheduling problems, this paper proposes MCMP-SRI and MCMP-SRSI, which are two multirecombination schemes that combine studs, random and seed immigrants. Evidence of the improved behaviour of the EAs when inserting problem-specific knowledge is provided. Experiments and results are discussed.