Abstract
Real-Time demand estimation is crucial for real-Time management of a drinking water system (DWS), which includes minimization of operating cost, emergency response, water quality maintenance, etc. On the other hand, real-Time demand estimation from a limited number of measurements is not possible without clustering consumer nodes assuming that the consumer nodes within a cluster have the same (or similar) temporal demand patterns. However, clustering nodes within a DWS model without consideration of the spatial distribution of measurement locations may make the demands unobservable within the estimation process. A machine learning based clustering algorithm, called the self-organizing map (SOM), is presented that clusters the nodes of a DWS model based on the sensitivities of the measurement locations to the nodal demands. The algorithm was applied to the Net3 network, an example network distributed with EPANET. The performance of the SOM based clustering was evaluated using data generated with a hypothetical cluster scenario representing unknown actual consumer distributions. Demand multipliers for the SOM clusters were estimated with a Bayesian approach. The SOM clusters showed good observability properties, and resulted in good representation between the simulated and synthetically generated flow measurements. However, the simulated flow measurements at unmonitored locations varied widely, which is an important consideration if adequate transport characteristics are of interest.
Original language | English (US) |
---|---|
State | Published - 2018 |
Externally published | Yes |
Event | 1st International Joint Conference in Water Distribution Systems Analysis and Computing and Control in the Water Industry, WDSA/CCWI 2018 - Kingston, Canada Duration: Jul 23 2018 → Jul 25 2018 |
Conference
Conference | 1st International Joint Conference in Water Distribution Systems Analysis and Computing and Control in the Water Industry, WDSA/CCWI 2018 |
---|---|
Country/Territory | Canada |
City | Kingston |
Period | 7/23/18 → 7/25/18 |
Keywords
- Demand
- Self-Organizing Map
- Water Distribution Network
ASJC Scopus subject areas
- Computer Science Applications
- Water Science and Technology