Comparison study of dispatching rules and heuristics for online scheduling of single machine scheduling problem with predicted release time jobs

Abstract

In real-life dynamic scheduling systems, the release times of some important and/or urgent jobs are usually known deterministically or predictably in advance. When making dynamic scheduling decisions, we are expected to get better scheduling solutions if these jobs are considered before they actually arrive. However, to the best of our knowledge, we have not yet found corresponding research to date. This paper addresses a dynamic single machine scheduling problem with predicted arriving jobs (DSMSP/PRJs). We give a detailed description and the corresponding mathematical programming model with the objective of minimizing total weighted tardiness (TWT). With respect to the inclusion strategy of PRJs, which is one of the key technologies affecting the solution quality, we propose a job-based inclusion (JBI) procedure that focuses on the influence on the tardiness of PRJs rather than on a certain length of lookahead window. The novel JBI procedure shows obvious superior performance to the time-based method in the literature with respect to TWT. The discrete event simulation based on dispatching rules/heuristics (RHs) is adopted to address the DSMSP/PRJs. Six weighted dispatching rules and two improved heuristics based on the DT-TD (Decision Theory-Tactically Delayed) are proposed. Large-scale and systematic experiments show that the proposed RHs perform well. Especially, the DT-TD2 heuristic has overwhelming superior performance to the others. For the huge volumes of experimental raw data produced by 1350 simulation runs under 135 problem scenarios, we innovatively put forward the overall procedure and methods of data processing and analysis oriented to the different purposes. The systematic data analyses show that, under the various problem scenarios, considering PRJs in advance can significantly improve the scheduling solutions as long as appropriate RHs are used. The analysis results also reveal that, in the scenarios with lower shop utilization, lower percentage of PRJ, higher weight of PRJ, and looser due date of normal job (NJ), it is more obvious in the improvement of solutions, implying that it is more necessary to consider PRJs in advance.