【摘要】 DMA是控制城市供水系统水量漏失的有效方法之一，而精准的水量预测是实现有效水量调度、漏损控制和资源优化配置的关键前提。针对城市混合用户DMA小区短期用水量预测提出了一种基于二维卷积改进的多变量时间序列长短期记忆神经网络水量预测算法，并和其他时序预测算法在MAE、MSE、RMSE、MAPE、SMAPE指标上进行实验对比分析，实验结果表明所述算法在混合用户DMA小区短期用水量预测上效果更好。更多还原

【Abstract】 DMA is one of the effective methods to control water leakage in urban water supply system, and accurate water volume prediction is the key premise to achieve effective water scheduling, leakage control and optimal allocation of resources. In this paper, a water consumption prediction algorithm based on two-dimensional convolution improved multivariable time series short-term and long-term memory neural network is proposed for the short-term water consumption prediction of urban mixed user DMA community, and compared with other time series prediction algorithms in Mae, MSE, RMSE, MAPE, smape indicators. The experimental results show that the algorithm in this paper has better effect on the short-term water consumption prediction of mixed user DMA community.更多还原