Applying data mining to data analysis in manufacturing

Abstract

In analyzing data from advanced manufacturing processes, it is important to integrate various types of data such as numerical, symbolic, and time series data, however, so large is the volume of data created by integration that engineers are not able to examine all of it. Data mining is a method for extracting information from large databases that can help to analyze the integrated data obtained from advanced manufacturing processes. We have developed a data mining method for analyzing manufacturing data that consists of three steps-feature extraction, combinatorial search, and presentation. We applied the method to LSI fault analysis and found that data mining is useful for indicating to engineers where to focus their attention when looking for faults.