Study Of Water Quality Prediction Model In Urban Water Distribution Systems

Water distribution system in the city takes the important responsibility of transmission and distribution of drinking water from water pant to the users. Anyway, because of the physical, chemical and biological agent in the pipeline, there is variable water quality frequently. Research the water quality variation law in the distribution system of water supply plant, establishing water quality model of distribution system, putting forward corresponding control measures and providing water supply plant reliable rationale to improve water supply technology and enhance its management, all of these have important real significance for the water quality.This paper made the water supply system as research object of one university in Tianjin, considering the practical situation, decided the experiment conceptual. Through the experiment monitoring to the network water quality, much data has been collected and prepared for establishing and checking the model.Partial least-squares Regression was used to established the statistic model of pipe network water quality, making turbidity and Fe as control indexes. Then PLS in SAS/STAT system was used to establish the statistic model and make the solution. The water quality model was established each monitoring spot in the demonstration plot and the turbidity and Fe in the pipe network have been predicted. The results indicate that the experiment value and analogue are anastomotic, and the accuracy of model is high, which can echo the variation of turbidity and Fe in the network. Only to the application effect, the model established by Partial least-squares Regression has perfect predicted effect for tow water quality indexes at the same time and can be used in practice. But the model has worse adoptability for the data variation, if the variation was severe, the predicted effect would be worse.At last, the paper promoted that strengthening pollution mechanics research, increasing parameters evaluation value accuracy, research the uncertainty of model and making GIS bonded to water ambient analogue, control and decision will be the important development direction of the water quality model.