Research On Multi-Objective Mid-Term Production Planning Optimization Model In Semiconductor Manufacturing

As one of the most fast developing and advanced industries, the semiconductor manufacturing industry leads and makes full use of the state-of-art achievements in science and technology. It's a high profit capital intensive industry which couples with high risk. Semiconductor manufacturing is a process with highly complex route and highly capital-intensive. Its industry market faces drastic changes with fleeting opportunities. As its development and growth, new industry features such as the wide varieties of products with shortened life cycles, the multi-factory, intricate working procedure, complicated BOM, expensive machines with long delivery cycle, and flexible manufacturing which requires higher utilization of machines come out continually. This change makes planning work in semiconductor manufacturing industry increasingly complex. Thus, the enterprises have to shorten the market reaction time, improve the production planning level, and consider a variety of relevant factors to make a more effective production plan in order to utilize the limited production resource, meet customer demand at a great extent, and maximize profits.An excellent enterprise production planning should fully take a wide variety of product characteristics into account, adapt to the flexible production resources and intricate supply chain of semiconductor manufacturing industry, as well as fast reaction to the change in order and production capacity. The traditional planning work which is basically solved through spreadsheets and heuristic approaches based on the local data, relies highly on the experience and wisdom of planners. It could not adapt this revolution. Enterprises are eager to find a kind of more accurate and more feasible technique which could handle with massive data to make a more scientific and reasonable product plan.In the thesis, the new features of semiconductor industry and the specific production process of semiconductor manufacturing are considered; the latest Advanced Planning and Scheduling (APS) theory is applied; a variety of constraints including the materials, capacity, resource utilization ratio, priority, inventory, throughput time (TPT), yield of semiconductor assembly and test manufactory are took roundly into account; the multiple periods material flows are modeled to transportation problem with transshipment; and the construction of mid-term production planning multi-objective mathematical model in semiconductor assembly and test manufactory is finished. Then, the model is solved by ILOG CPLEX solver utilizing Linear Programming (LP) based approach, and referencing Goal Programming (GP) ideology. In addition, an experiment is carried out based on the actual data of a large-scale semiconductor assembly and test manufactory. The result of practical experiment in factory proves that the production efficiency and response time are greatly improved after the launch of the production planning model proposed in this thesis while the redundant inventories and unmet demands have been largely reduced. The thesis provides a valuable solution for solving practical production planning issues. At last, a conclusion is made and the limitations as well as the further research directions are discussed in this thesis.