Automatic classification for mining process operational data

In process-plant operation and control, modern distributed control and automatic data logging systems create large volumes of data that contain valuable information about normal and abnormal operations, significant disturbances, and changes in operational and control strategies. These data have tended to be underexploited for a variety of reasons, including the large volume and lack of effective automatic computer-based support tools. This paper considers a data mining system that is able to automatically cluster the data into classes corresponding to various operational modes and thereby provide some structure for analysis of behavioral responses. The method is illustrated by reference to a case study of a refinery fluid catalytic cracking process.