Dispatching rule extraction method for job shop scheduling problem

Abstract

A novel decision tree based approach was proposed to extract new dispatching rules from optimal schedules generated by traditional optimization methods for job shop scheduling problems(JSP).Timed Petri net was used to formulate JSP and an improved Petri net based branch and bound algorithm was developed to generate solutions.On this basis,a novel dispatching rule extraction method was proposed to mine scheduling knowledge implicit in solutions.The extracted knowledge,formulated by a decision tree,was used as a new dispatching rule.The new dispatching rule constructed by the proposed method was tested on a set of test cases and benchmark problems for the objective of minimizing makespan of JSP.The results show that the schedules generated by the extracted rule have smaller makespans than the schedules generated by the existing dispatching rules extracted from solutions of JSP and the common priority dispatching rules,verifying the feasibility and validity of the proposed method.