Stochastic, multi-objective job-shop scheduling model
Abstract
A stochastic, multi-objective job shop production scheduling model is developed in this research. The prediction of the impact of future scheduling decisions becomes uncertain because of the high integration among the scheduling constraints and the uncertainty of uncontrollable and unpredictable facts. Uncertainties from different sources in the production environment are taken into consideration. Uncontrollable and unpredictable facts include changes in order priorities, processing times, machine availability, delivery delays in raw materials, random yields, defective parts, and so forth. In order to capture the uncertainty in job shop environments, uncertainty is put explicitly into the scheduling model through the representation of the job and machine idle times as probability density functions. A set of heuristics is developed to promote a solution strategy that requires less time and effort than traditional approaches. The heuristics are developed to offset job and machine duration variances. They promote useful properties to direct the search towards the prioritization of scheduling alternatives that dynamically seem to promote the minimization of overall idle times. The strategy for the prioritization of the scheduling alternatives is based upon the maximization of the probability of minimizing overall idle times. An evaluation function is designed to combine and evaluate the expected contributions each alternative's job and machine offers. To evaluate the performance of the model, several well known job-shop problems were tested, and the makespan obtained by the model was compared with the optimal solutions.
This paper has been withdrawn.