The effects of a machine failure on the robustness of job shop systems-the predictive-reactive approach

The paper presents the classification of methods of dealing with uncertainty in scheduling problems. Two proactive approaches are identified: predictive-reactive (proactive with prediction) and proactive-reactive (proactive without prediction). In the first approach, researchers use prediction methods to predict maintenance time. Next, the influence of a disturbance on the schedule using the robustness measures is examined. In the second approach, the proactive schedule is achieved for the best sequence of idle times between jobs or batches taking the advantage of the simulation process. This paper presents the results of predictive-reactive approaches using both: the Hybrid Multi-Objective Immune Algorithm and heuristics based on priority rules. Two predictive heuristics are proposed in order to generate robust and stable schedules under uncertainity. The heuristics differ in the scheduling procedures applied for less flexible operations that are predicted to be disrupted by a machine failure. Computer simulations are conducted for job shop systems.