Applying Data Mining Techniques for Process Parameter Optimization in Casting Industry

• Main Authors: Dong, Ya-Yu, 董雅瑜
• Other Authors: Su, Chao-Ton
• Format: Others
• Language: zh-TW
• Published: 2018
• Online Access: http://ndltd.ncl.edu.tw/handle/cv3gg3

碩士 === 國立清華大學 === 工業工程與工程管理學系所 === 106 === With the advancement of manufacturing technology and the flourishing development of information technology, casting industry is faced with the increasingly competitive market, so companies must enhance their product’s quality and reduce manufacturing costs, and clarify what highly influences the process to have the key competitive advantage. However, simply relying on domain knowledge or rules of thumb is unable to identify the root causes of quality problems effectively. This study applies data mining techniques for the process improvement issue of casting industry and proposes a general procedure for attribute selection. Five data mining techniques, including artificial neural network (ANN), random forest (RF), support vector machine (SVM), rough set theory (RST), and regression analysis are used to select the important attributes. This study aggregates the results from each method to identify the key parameters and builds the reduced model. In the end, the artificial neural network and genetic algorithm (GA) are utilized for optimizing the selected process parameters.The proposed procedure was employed to analyze the manufacturing data of a casting company in Taiwan. The research results presented that nine key process parameters were identified from seventeen original attributes and then the optimal combination of key parameters was obtained. In addition, the reduced model still maintained the exceptional ability to perform adequately, which confirmed the feasibility of the proposed attribute screening procedure.