Research of water volume prediction method based on Transformer deep learning
The prediction of water supply is the basis and foundation for guiding the intelligent dispatching and refined management of urban water supply network.Among them,DMA water supply prediction provides support for scientific management of regional leakage and secondary wa-ter supply.Aiming at the problem that the accuracy of DMA water supply prediction is insufficient at present,a water supply prediction method based on deep learning is proposed.Based on the Transformer architecture in deep learning,a Transformer time series prediction model is built which using self attention mechanism to learn the dynamic change mode of water quantity series,to carry out relevant research on short-term water supply prediction.The model is applied to a sec-ondary area and a tertiary DMA area under the water supply zoning metering system of Hefei City to predict the short-term water supply at the regional entrance for 1h.The prediction results were compared with the results of the long short memory network(LSTM)model and the support vec-tor regression(SVR)model to verify the accuracy of the model.
Water demand forecastingMachine learningTime seriesDMA
万方数据
维普
