Study Of Bayesian Probabilistic Forecasting Method For Urban Daily Water Consumption

With the acceleration of the urbanization, the size and scope ofurban water distribution networks are fast increasing, and the networks ofpipelines are becoming more and more complex. The traditionaloperation for urban water supply system faces challenge and optimaloperation for water supply system is imperative under this situation.Urban daily water consumption forecasting is the key to optimalmanagement of water supply system, and its accuracy determines whetherthe optimal operation scheme can be applied to the real water distributionsystems or not.Research progress of urban daily water consumption forecasting issystematically summarized at home and abroad in this paper, and themain work of this paper is below:(1)The correlation of the daily water consumption series wasanalyzed by the mutual information method, and the influence factorsaffecting daily consumption were analyzed by a rough set algorithm ofweighting-coefficient cumulative estimation. A daily water consumptionforecasting model based on the support vector machine (SVM) was developed.(2) The Bayesian theory was introduced to build the probabilisticforecasting model. In the model, the prior density and likelihood functionbased on SVM were built. The Bayesian probabilistic forecasting modelwas solved by the adaptive Metropolis algorithm, and the probabilisticforecasting for daily water consumption was obtained. Case study showsthe proposed method has better estimating performance, and it providesthe confidence interval of daily water consumption forecasting for thescientific and reliable decision-making of the optimal dispatch of waterdistribution network.