Online Classification of Abnormal Situations in Operation of Industrial Processes Based on Alarms and Process Variables
alarm processing, failures diagnosis, industrial automation, alarm systems, intelligent systems.
Industrial processes are subject to failures in their thousands of components at any time and can lead to shutdowns, loss of product quality, equipment damage or even accidents. In this sense, the alarm system is necessary to aid in the identification of process abnormalities. However, during a process failure it is common for the operator to be subjected to hundreds of alarms causing overload beyond the human processing capacity. This phenomenon is known as alarm flood and to treat them properly is a challenge for the modern alarms systems. Thus, the present work aims at the development of an on-line alarm processing methodology capable of assisting the operator in the identification and classification of abnormal situations of the process, especially in moments of alarm overload. To validate the proposal, a case study was carried out on a process simulator widely used and accepted by the scientific community called Tennessee Eastman Process.