Research On Optimal Operation Of Municipal Water Supply System

Main contents for optimal operation of municipal water supply system was introduced, in allusion to status quo and existing problem of optimal operation model, the model of water prediction, state model of water supply system and optimal operation model were delved into.Because hourly water demand is affected by weather, season, activities of dwellers and the condition of industrial production, it has the characteristics of periodicity and trend. Since traditional experience nonlinearity learning methods, such as Neural Network, have the defect of local minima, Support Vector Machine (SVM) arithmetic was introduced which has structural risk minimization rule. SVMdark was adopted to optimize parameters of SVM's core function, so that the time of prediction was shortened and prediction accuracy of hourly water demand peak was improved. According to the hourly water demand forecasting results of one Economic Development Area (EDA) of North China, the rationality and effectiveness of this arithmetic was proved.Water supply system is a complicatedly dynamic nonlinear system, it is difficult to simulate by using one or several explicit functions in pipe networks. Considering macroscopical model of water supply system not only can depict the status of the whole pipe networks, but also the cost of modeling is low, and less time is took, a pressure macroscopical model was proposed which was based on SVM arithmetic. The reasonability of this model was embodied in the practice of EDA pipe networks. According to the use of pressure macroscopical model in EDA pipe networks, the rationality and effectiveness of this model was proved, and the building of macroscopical model of optimal operation of municipal water system was solved.The direct optimal model of multi-objective mixed discrete variables for water supply system was established, Methods to deal with each constraints were put forward by replacing complicated network hydraulic equation with pressure macroscopical model. The Lingo software was adopted which solve the nonlinear model in order to globally optimize the model of multi-objective mixed discrete variables, so as to solve effectively the direct optimal model of water supply system which has more variables and constraints. The reasonability and effectiveness of this arithmetic was proved by real examples.