Process Data Analytics in Upstream Processes

Systems Conversation with Biao Huang, University of Alberta

Modern industries are awash with large amounts of data. Extraction of information and knowledge discovery from data, particularly from day by day routine operating data, is especially challenging. There exist numerous challenging issues such as data nonlinearity, non-Gaussian distributions, high dimensionality, collinearity, multiple modal operations, outlying data points, missed measurement, etc., that must be considered during the information extraction process. These challenges are especially true for upstream processes where systems are more complicated and sensors are less reliable. This presentation will discuss state-of-the-art development of data analytics for oil sands processes with focus on image processing and data fusion with industrial applications in oil-water interface detection.