Sensor fault detection for urban drainage systems using redundant measurements

Abstract

ABSTRACT In urban drainage systems, sensors are used for process monitoring, maintenance, planning, and control. The safety, reliability and performance of the systems are therefore largely dependent on the accuracy and reliability of the sensors. Faulty sensors may lead to poor operational performances in terms of flood and pollution control. An innovative real-time sensor fault detection approach based on the statistical properties of redundant measurements is proposed herein to improve the operation and maintenance of urban drainage systems. This approach is easy to implement, does not require estimates of the process variables and can be configured to detect a large set of anomalies, including incipient failures. The Louisville and Jefferson County Metropolitan Sewer District’s urban drainage system in Kentucky, USA, illustrates the performance of the algorithm.