期刊名称：International Journal of Distributed Sensor Networks
印刷版ISSN：1550-1329
电子版ISSN：1550-1477
出版年度：2019
卷号：15
期号：2
页码：1
DOI：10.1177/1550147719832802
出版社：Hindawi Publishing Corporation
摘要：Surface mount technology is an important process in modern electronic circuit manufacturing. Quality control problems have arisen in this area because of the increased design and processing complexity of electronic circuits. Identifying the cause of a fault shortly after its occurrence is critical; however, human fault analysis is inaccurate and time-consuming. Here, we propose a data analysis method that provides actionable information that can easily be interpreted to facilitate rapid identification of fault cause in surface mount technology. The proposed method divides each input variable into a certain number of partitions, and then, the proportion of faults in a partition is calculated in comparison to the proportion of faults in the entire data set. The analytical results are provided to the user with a list that includes the fault causes and a corresponding density histogram for visualization. Real-world surface mount technology data were employed for a case study, in which raw data were preprocessed into an integrated data set consisting of 14,847 rows and 12,929 columns. The proposed method showed reasonable results in approximately 65 s, and the visualization of the results provided a suitable basis for intuitive interpretation, thus demonstrating the method’s ability to generate an efficient analysis in a practical application.
关键词：Data mining; surface mount technology; fault analysis; fault cause identification; smart manufacturing
