Application of data mining technologyin quality control-Chungli Industrial Zone in the case example of the company's PCB Test Day Report

• Main Authors: Xian-Cong Chen, 陳賢聰
• Other Authors: Mei-Chun Chen Ph.D
• Format: Others
• Language: zh-TW
• Published: 2011
• Online Access: http://ndltd.ncl.edu.tw/handle/38718405536349741026

碩士 === 萬能科技大學 === 資訊管理研究所 === 99 === PCB (Printed Circuit Board, PCB) in the manufacturing process, finished goods tend to be people, machines, materials, environment, process, and so the quality factors causing yield variability, past quality control, led mainly applied statistics, China's industrial development accounted for a significant contribution to the position. But today's increasingly sophisticated management techniques, management methods and computer performance has matured with each passing day, if application of database technology within the report data entry on computer and data processing by computer algorithms that can provide decision support for management. This study data collected in this study is already the end of the process, the next process for the product packaging, the study could not collect the whole process to process information, so the integrity of research needs of other researchers to gather together. In this study, analysis of test data sample size for the 1111 samples, the sample period from May 2009 to April 2010. Use a decision tree algorithm with the two association rules, the case provides individual test day report of PCB input to Excel spreadsheets, use mining tools SPSS PASW Modeler 13.0 Chinese trial version, open the mine after the tool to import Excel data analysis. The results of this study is able to enter the machine at the time of each test period yield on the quality of output to do a full show, but also for machine analysis of the optimal parameter set basis. Enter the testers in the time period yield on the quality of output, can understand the optimal portfolio mix between testers and machine (machine parameters). This study only take a certain point in time during the data analysis test time and test the continuity of data types such as date, start the analysis does not analyze within a certain time and date the relationship between yield and quality, the other will require continuous data type into a class-based data type, and content within the set threshold conditions required for the results of the last to appear in the tree graph with the results of association rules. Future research suggestions to collect research data if the longer time period of data (longitudinal data), the more managers can reference value.