Case Study: Parameters Optimization Process for Flat Panel Display Manufacture

Taiwan's flat panel display manufacturing is famous internationally. Its outstanding manufacturing capability can produce high quality panels in the shortest time. In the production process of flat panel displays, the process yield is the best indicator reflecting the good and bad of manufacturers. The yield is direct. The decision is to the company's profit and reputation. Add to this, it is also the trust and acceptance of the company's customers. When the products on the production line are abnormal in quality, the process and equipment engineers must immediately discuss the crux of the problem and improve it. For decomposition of analysis and process management to find out the method, it needs to carry out the problem in linear verification of confirmation whether the problem is a single problem or not. Furthermore, it will produce customer complaints, dissatisfaction with the company, causing the company's losses. Therefore, using Knowledge Discovery from Data (KDD) to establish a process of big data analysis is core technology in our work. We proposed the improvement process that how to quickly find the variation factor and couple with machine learning. Hence the research is based on data pre-processing, modeling. The experiment shows that we find out the right parameters and optimze them. Finally, the yield improvement was increasing to 66%.