基于Transformer深度学习的供水量预测研究

Research of water volume prediction method based on Transoformer deep learning

供水量预测是指导城镇供水管网智能化调度和精细化管理的依据与基础,针对目前独立计量区域供水量预测准确性不足的问题,提出一种利用深度学习的水量预测方法,基于深度学习中的Transformer架构,搭建Transformer时间序列预测模型,利用自注意力机制学习水量序列的动态变化模式,开展短期供水量预测的相关研究。将模型应用于合肥市供水分区计量体系下某二级区域及某三级小区DMA,进行区域入口1h短期供水量预测。对比使用长短时记忆网络模型及支持向量回归模型,结果表明,Transformer水量预测模型的预测效果最优,水量预测模型的预测精准度获得有效提升。

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 water 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 secondary 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 lh.The prediction results were compared with the results of the long short memory network(LSTM)model and the support vector regression(SVR)model to verify the accuracy of the model.