In many sectors of commercial operation, the scheduling of workflows and the allocation of resources at an optimum time is critical; for effective and efficient operation. The high degree of complexity of a “Job Shop” manufacturing environment, with sequencing of many parallel orders, and allocation of resources within multi-objective operational criteria, has been subject to several research studies. In this thesis, a scheduling system for optimizing multi-objective job shop scheduling problems was developed in order to satisfy different production system requirements. The developed system incorporated three different factors; setup times, alternative machines and release dates, into one model. These three factors were considered after a survey study of multiobjective job shop scheduling problems. In order to solve the multi-objective job shop scheduling problems, a combination of genetic algorithm and a modified version of a very recent and computationally efficient approach to non-dominated sorting solutions, called “efficient non-dominated sort using the backward pass sequential strategy”, was applied. In the proposed genetic algorithm, an operation based representation was designed in the matrix form, which can preserve features of the parent after the crossover operator without repairing the solution. The proposed efficient non-dominated sort using the backward pass sequential strategy was employed to determine the front, to which each solution belongs. The proposed system was tested and validated with 20 benchmark problems after they have been modified. The experimental results show that the proposed system was effective and efficient to solve multi-objective job shop scheduling problems in terms of solution quality.
|Publication status||Accepted/In press - Jun 2016|