Research On Methods Of Simulation Test And Result Evaluation For Dispatching Rule-based Algorithm In Dynamic Job Shop Scheduling Problem

Most actual scheduling of manufacturing system are complex, dynamic andlarge-scale scheduling problems. For this type of scheduling problems, using thealgorithm of scheduling rules, it can be solved quickly and efficiently to get the closeoptimal solution or accurate solution in order to meet the engineering requirements. Inthis paper, methods of simulation test and result evaluation are studied for dispatchingrule-based algorithm in dynamic job shop scheduling problem.For the three kinds of typical dynamic job shop scheduling problems：Dynamic jobshop scheduling problem with jobs release one-by-one intermittently，dynamic job shopscheduling problem with jobs release batch-by-batch intermittently and dynamic jobshop scheduling problem with machine breakdown．This paper presents the methods togenerate the basic attributes of each element of scheduling problem, then establishes themathematical model, and studies the technique of case generation, besides, the data isobtained through the simulation cases. This paper also provides the design methods ofthe simulation scheduling test, which include the method to decide the total number ofmachine and the total number of jobs, methods and steps of acquiring sample data ofsimulation scheduling, methods of how to set the shop utilization and the improvedmethod of setting comfortable degree coefficient of due date. It studies the methodwhich combines the single factor-completely randomized block design, the single factorvariance analysis and simulation experiment data statistics and evaluation of thescheduling rules while using the scheduling rules as the independent variable. Finally,based on two job shop problems: dynamic job shop scheduling problem with jobsrelease one-by-one intermittently and dynamic job shop scheduling problem with jobsrelease batch-by-batch intermittently, this paper uses heuristic algorithm based onpriority rules to conduct a complete scheduling simulation experiment, then verifies thefeasibility and effectiveness about the case generation, simulation scheduling, datacollection and evaluation of the results mentioned above.