인공신경망을 이용한 상수관망의 에너지 손실 예측

Abstract

Supervision of a stable water supply should be supported by proper analysis of the levels of energy supplied and consumed. Accurate prediction of the energy lost to friction in pipelines can be helpful for effective management of a water distribution system. The present study applied an artificial neural network (ANN) in predicting energy lost due to friction in a drinking water distribution system. The study site was the Cheongna water distribution system in Incheon, and the results obtained through the EPANET model were used to estimate energy lost to friction. Major hydraulic factors of the pipelines were used as input data for predicting the coefficient of friction loss, friction loss head, and energy loss through the ANN. The findings were compared to the model simulation values to determine the applicability of ANN. The findings in the study showed that using an ANN produced highly accurate estimates of coefficient of friction loss and friction loss head in the pipelines, whereas energy loss could not be accurately estimated. This study suggest that an ANN can be used to estimate factors related to friction in pipelines.