본문 바로가기
  • Home

Estimating Non-revenue Water Ratio Using ANN based on PCA with Data Normalization in Water Distribution Systems

  • Crisisonomy
  • Abbr : KRCEM
  • 2018, 14(3), pp.103-118
  • DOI : 10.14251/crisisonomy.2018.14.3.103
  • Publisher : Crisis and Emergency Management: Theory and Praxis
  • Research Area : Social Science > Public Policy > Public Policy in general
  • Received : January 23, 2018
  • Accepted : March 13, 2018
  • Published : March 31, 2018

Jang, Dong Woo 1

1인천대학교

Accredited

ABSTRACT

The non-revenue water (NRW) ratio in water distribution systems is an index of the loss of water supply caused by pipe burst, operational loss and physical factors. NRW ratio is a comparative index for city, province and DMA (district metering area) in the domestic water supply maintenance project. An investigation of the factors affecting the NRW as well as its estimation have become increasingly important in an economic sense. In this study, PCA (principal component analysis) and ANN (artificial neural network) are used as statistical methods to estimate the NRW ratio. The normalized data were obtained through the Z-score method, and then the PCA-ANN model was constructed for the NRW ratio estimation. Accuracy assessment was performed to compare the observed NRW ratio with the estimated ratio from the ANN model. The results show that the PCA-ANN model is more accurate than the single ANN and the estimation results differ by the number of neurons in the hidden layer of ANN. As for the six independent variables used in this study, the accuracy of NRW ratio prediction was found highest when 12 neurons were used.

Citation status

* References for papers published after 2022 are currently being built.