Author(s): G. Riss; M. Romano; K. Woodward; F. A. Memon; Z. Kapelan
Linked Author(s):
Keywords: Confusion matrices; Event recognition; Online monitoring; Water treatment works
Abstract: This paper presents a study aimed at improving the event detection capabilities of the existing, threshold-based detection system at a selected Water Treatment Works operated by United Utilities. The study shows that improvements can be achieved by using optimised threshold and persistence values identified by performing a sensitivity type analysis. The main findings from this study show that, although an overall increase in the true detection rate and decrease in the number of false alarms can be achieved, the high number of false alarms remains an issue. To address this problem a new event detection system based on suitable relations across multiple signals will be developed as part of future work.
Year: 2018