Categorical data analysis for equipment failure prediction

The effective approach to monitor the health state of equipment has long been a concern of industrial applications. This paper focuses on the categorical data analysis to build equipment degradation model for predicting equipment failure and monitoring the health state of equipment. Since large volumes of categorical data are generated and collected in most manufacturing execution systems, how to extract useful knowledge from the data become more and more compelling. Major issues in this area have been identified through company visits, data collection and processing, and case study. Evaluation and development of ant colony and Pareto profile based techniques, for model of equipment performance degradation based on categorical data analysis are presented. Through the case study on a test-bed company, we demonstrated the feasibilities of the proposed technologies for performance degradation modeling for complex equipment.