ABSTRACT: With the growing demands placed on industrial production to achieve an integrated smart plant, factories require the development of efficient and viable methods for analysing and assessing the available data. Recently, the emphasis has focused on the development of data-driven control methods that use economic key performance indices (KPIs). These methods seek to determine which of the observed changes in the plant performance have the largest impact on the economic performance of the plant and then take appropriate control actions to maintain the overall plant performance. The advantage of such an approach is that it allows for an effective prioritisation of different problems. However, it requires the availability of reliable measurements at the desired sampling rates. In practice, the required variables, for example, in the oil sands with bitumen concentration or the hot steel mill rolling processes with steel roll thickness, are often not available at the desired sampling rates. In order to obtain these values, it is necessary to incorporate soft sensors that can provide estimates of the variables and, hence, overcome this lack of information. Extensive tests of these soft sensors to detect faults show that it can improve the precision and accuracy of the fault detection system. Future work needs to be performed to deal with uncertain, unknown, time delays and sampling instances.
Dr. Yuri A.W. Shardt is currently a Research Assistant at the University of Duisburg-Essen in Duisburg, Germany. Between September 2013 and July 2015, he held the prestigious Alexander von Humboldt Fellowship at the same university. During his sojourn at the University of Duisburg-Essen, he has been working on the problem of the development of soft sensors for fault-tolerant control. Furthermore, he is interested in the development of holistic control systems that can control a plant from start-up to shut-down. He obtained both his Doctorate in 2012 and Bachelor of Science in Chemical Engineering (Computer Process Control Option) in 2008 from the University of Alberta. He has written one book just published by Springer, called Statistics for Chemical and Process Engineers. He also has many journal articles and conference presentations.