A new method for minimizing cell underutilization in the process of dynamic cell forming and scheduling using artificial neural networks

Abstract

Cell-load variation is considered as a major shortcoming in cellular manufacturing systems. It can cause long queues in front of machines and impose extra costs to the cellular layouts. In this paper the impact of inflation on cell-load variation in cellular manufacturing systems is examined. For this purpose, a new method is proposed for scheduling dynamic cellular manufacturing systems in the presence of bottleneck and parallel machines. The aim is finding the trade-off values between in-house manufacturing and using outsource services while system costs are not deterministic and may be varied from period to period by inflation. To solve the model, a hybrid Multi-layer perceptron is developed because of the high potential of outcomes to be trapped in the local optima. Our findings show that the condition of dynamic costs affects the routing of materials in process and may induce machine-load variation that yield to cell-load diversity. An increase in changing costs causes the loading level of each cell to vary, which in turn results in the development of “complex dummy sub-cells.” The results indicate that the proposed method can significantly reduce the level of cell-load variation in CMS.

Keywords

• facilities planning
• cellular manufacturing systems
• cell scheduling
• artificial neural networks