Research On Modelling And Optimazation On Production Planning In Semiconductor Back- End Manufacturing

The semiconductor manufacturing industry is a typical industry of capital-intensive. The processing equipment which are located in the production line are very expensive. Besides the market demands are highly uncertainty, but quick response for customers is a must. The production of semiconductor manufacturing is based on the market prediction which is produced before the customers’orders, so a reasonable and executable production plan is a key link for the semiconductor industry to make full use of the factory’s resources. Master production scheduling system is a very important tool for the production planning of manufacturing company which is based on the market prediction and capacity constraints. This thesis takes the semiconductor back-end production line as the research background to set up the mathematical model for the planning system based on the features of semiconductor industry.First, this thesis has analyzed the business rules for semiconductor manufacturing such as Binsplit, inventory strategy, freeze period and so on. Two planning models for traditional production line and outsourcing production line based on these business rules has been setup. For the traditional production line planning model, it is constraint by the capacity, demand, and inventory with multiple objectives. The thesis has solved the model with ILOG by four different methods, penalty method, linear method, Satisfaction method based on subordinating degree function and layered sequence method. By comparison, it proves that the model works and another way provide a better way to solve multi-objective planning model. Moreover, this thesis has research on the outsourcing line for planning model for different supplier market. For equal and master-slave relationship, this thesis use the Gounod game model and Stanley game model to analyze their different output decision-making. The semiconductor manufacturer allocate the orders according to this analysis and make it a constraint in the planning model and solve the model with a case as well. Last, this thesis summarize the conclusions.