Study on On-Line Prediction of Node Water Demands and Optimal Water pressure Control in Water Supply Network

(1)给水管网节点需水量在线预测方法。为了实现城市供水管网的优化控制，需要对城市供水管网的需水量进行准确预测。因此，本研究提出一种可即时预测城市给水管网节点需水量、节点水压及管道流量的方法。该预测算法基于卡尔曼滤波理论，将水头损失的连续方程和线性化方程转化为观测方程。在预测算法中，关于网络中的一些管道流量、节点液压和阀门开度的传感器信息是可观测变量。为了证明该方法的准确性，它被应用到福冈市供水网络使用综合生成的和实际的节点用水需求。此外，该方法的基本特点，如改变阀门开度的预测精度的影响进行了详细的研究。(2)考虑阀门水头损失的节点能量法管网分析。为了控制管网中的液压，有必要首先分析网络流量，考虑阀门引起的水头损失以及摩擦。在这项研究中，管网分析的节点能量法制定，其中考虑阀门的水头损失，而不增加节点的数量所造成的设置额外的阀门。文中还编制了计算程序，可方便地应用于有阀和无阀的情况。实际管网的应用结果表明了该方法和程序的有效性。

(1) On-line prediction method of node water demandsinwatersupply network. To optimally control city water supply network, accurate predictions of water demands are required. Hence, this study presents a method to predict on-line the node water demands, node hydraulic pressures and pipe discharges in city water supply network. This prediction algorithm is developed based on the Kalman filtering theory, where the equation of continuity and linearized equation of head loss are transformed into observation equation. Sensor information on some of the pipe discharges, node hydraulic pressures and valve openings in the network are the observable variables in the prediction algorithm. To demonstrate the accuracy of the method, it is applied to the Fukuoka City water supply network using synthetically generated and actual node water demands. Moreover, the basic characteristics of the method such as the effect of changing valve openings on the accuracy of the prediction are investigated in detail.(2) Pipe network analysis by nodal energy method considering valve head loss. To control the hydraulic pressures in a pipe network, it is necessary to analyze first the network flow considering the head losses caused by valves as well as frictions. In this study, a pipe network analysis by nodal energy method is formulated, in which the valve head loss is considered without increasing the number of nodal points caused by setting up additional valves. The computer program for this analysis is presented, and it can be applied easily to the case of valves existing and to the case of valves not existing. Results of the application to the actual pipe network show the effectiveness of the method and program.