Comparison between fixed thresholds and genetic algorithm methods for water quality event detection

This study compares two methods of identifying contamination events in water distribution systems. The GA dynamic threshold versus the fixed threshold method which both result in tools designed to provide the decision maker with a comprehensive yet understandable view point of all data available. Results of the comparison show that the GA approach performs better than the fixed threshold method in the identification of contamination events and that the fixed threshold mechanism results in lower false alarms. The sensitivity of the GA is the result of the dynamics of the threshold, while the lower number of false alarms in the fixed thresholds model are due to the constant values of the thresholds. Some pre-determined customized parameters must be set in order for the GA to reflect the utilities view of the water distribution system characteristics.