Identifying Leakage Likelihood Using State Estimation and Bad Data Identification Methods

As the number and type of sensor deployments on water distribution networks (WDNs) increases, there is an opportunity to use the sensor information to improve the management and operation of the network.

Several techniques have been proposed in the literature to exploit data coming from pressure/flow sensors in order to provide an initial guess for the location of the leaks within a water network, thus reducing the time required by physical exploration. Most of the proposed methods are based on the analysis of residuals between the sensor data and an estimate calculated from prior knowledge of the nodal demands (Vento, 2009; Perez et al., 2010; Gertler et al., 2010). An integration of the residual analysis with state estimation, where the demands are estimated from the sensor data, was also proposed in (Andersen et al., 2000; Fusco et al., 2012). In (Wu, 2008; Wu et al., 2010), an optimization-based technique was proposed, where the leakage at the nodes is estimated by minimising the residuals between sensor data and model prediction, using genetic algorithms. Existing techniques, however, are not reliable in practical setups characterized by sparse sensors and may produce misleading diagnosis, with high rates of false positives or false negatives. Even though more sensors are going in all the time, in fact, the number of sensors is still usually small compared to the number of nodes in a skeletonised system model.

In this paper, tools based on state estimation and bad data analysis, popular in the power systems industry, are combined with factor analysis into a new method for detecting flow anomalies in water systems. A key feature of the new technique is an aggregation scheme whereby detected anomalies are mapped to a sub-graph of the network consistent with the density of sensors. This mapping contrasts with reporting anomalies for a single node when such resolution is not justified by the measurement density. The size of the flow anomaly is estimated along with its uncertainty to support decisions on possible corrective actions.

The proposed methodology, along with some background on state estimation and bad data analysis, is detailed in section 2. Both real and semi-synthetic results, on a real municipal DMA, are presented in section 3. Final comments are then given in section 4.

By: F, Fusco, B. Eck, S. McKenna

Published in: RC25445 in 2014

LIMITED DISTRIBUTION NOTICE:

This Research Report is available. This report has been submitted for publication outside of IBM and will probably be copyrighted if accepted for publication. It has been issued as a Research Report for early dissemination of its contents. In view of the transfer of copyright to the outside publisher, its distribution outside of IBM prior to publication should be limited to peer communications and specific requests. After outside publication, requests should be filled only by reprints or legally obtained copies of the article (e.g., payment of royalties). I have read and understand this notice and am a member of the scientific community outside or inside of IBM seeking a single copy only.

rc25445.pdf

Questions about this service can be mailed to reports@us.ibm.com .