Case-Based Reasoning On Capacity Planning In Semiconductor Manufacturing

Capacity planning in semiconductor manufacturing is one of hot research topics in industrial community. Firstly, Semiconductor manufacturing belongs to capital-intensive industry. Those semiconductor enterprises face tremendous risks because of expensive equipment, workshop and its long delivery time. Secondly, with continuously shorter semiconductor product life cycle and decreasing product profit, the competition among semiconductor enterprises is becoming fiercer day by day. Capacity planning in semiconductor manufacturing is also a very complicated problem. Its complexity is mainly due to multiple product mix, complicated manufacturing process, and requirement for high machine utilization. In addition, sensitivity of capacity to mix production and the uncertain demand makes it very difficult to determine equipment quantity to meet the forecasted demands.The problem of capacity planning in semiconductor manufacturing system is a typical multi-objective problem with characteristics of multi-restriction, plenty of variables and uncertainty. The existing optimization method is somewhat deficient either in theoretical analysis or practical application, which need to be studied deeply and systematically. Based on some simplifying assumptions, a multi-objective optimization model which considers maximizing profit, utilization of capacity and minimizing unmet demand is proposed. This paper use case-based reasoning (CBR), combined with adaptive resonance theory (ART) and multi-objective evolutionary algorithm (MOEA), to solve the problem. The CBR system is implemented by C++, applying improved ART-2 neural network to case retrieving and NSGA-II algorithm to case revising. Finally, an illustration shows how these techniques work effectively in solving the problem.