A semantic similarity based dispatching rule selection system for job shop scheduling with multiple production objectives

Job shop scheduling is an important activity which properly assigns production jobs to different manufacturing resources before production starts. Compared to other scheduling approaches that use optimal branch and bound algorithms, metaheuristics, etc., the dispatching rule based approach has been widely used in the industry because it is easier to implement, and it yields reasonable solutions within a very short computation time. The dispatching rule based approach uses a selected single dispatching rule (e.g. Shortest Processing Time or Earliest Due Date) or a rule combination depending on the current production objective like maximizing productivity, minimizing makespan or meeting the due dates. However, a dispatching rule or a pre-set rule combination always pursues a single and fixed production objective. This characteristic confines the flexibility of the scheduling system in practice. In order to address this issue, this paper proposes a semantic similarity based dispatching rule selection system that can achieve the intelligent selection of dispatching rules based on the user selected one or more production objectives for job shop scheduling. The intelligent selection is addressed by measuring the semantic similarities (based on ontology) between the user selected production objectives and the characteristics of the dispatching rules. The rule combinations will then be constructed by combining individual dispatching rules with similarity value based weights. A proof-of-concept demo has also been provided as a case study in this paper.