Data Analytics Applied to Reduce Hazards during Process Transitions

A data analysis approach based on machine learning is used to identify and characterize process data with standard transition processes. Transitions are a ubiquitous part of any continuous process occurring every time a valve is opened manually, a pump is started, or equipment is taken in or out of service. A data analysis framework is proposed to identify recurring transitions and detect deviations from normal transitions due to faults. We use classification, regression, clustering, dimensionality reduction, and other machine learning techniques to extract information from process data. The information extracted can then be used to identify clusters of data and detect similar patterns and be used as a basis to characterize the transition procedure. It is also possible to detect different trend relationships and characterize the trend responses.