Trajectory Clustering Approach for Reducing Water Quality Event False Alarms

Event Detection Systems (EDS) performance is hindered by false alarms that cause unnecessary resource expenditure by the utility and undermine confidence in the EDS operation. Changes in water quality due to operational changes in the utility hydraulics can cause a significant number of false alarms. These changes may occur daily and each instance produces similar changes in the multivariate water quality pattern. Recognizing that patterns of water quality change must be identified, we adapt trajectory clustering as a means of classifying these multivariate patterns. We develop a general approach for dealing with changes in utility operations that impact water quality. This approach uses historical data water quality data from the utility to identify recurring patterns and retains those patterns in a library that can be accessed during online operation. We have implemented this pattern matching capability within CANARY and describe several example applications that demonstrate a decrease in false alarms.