Study Of Water Quality Statistical Model In Water Distribution Systems

Water quality although acceptable when it leaves the treatment plant may deterioratebefore it reaches the user through drinking water distribution network. The purpose ofthis paper is to guarantee the security of water quality when it runs through waterdistribution network. The paper takes the water distribution system of a Tianjin collegeand a north-china water distribution network as objects. A systemic and profound workon the foundation and calibration of water quality model is done in this paper.A water quality-measuring scheme is formulated, associating the practice of waterdistribution system of the demonstration areas. Water quality information is obtainedfrom on-site monitoring and library measuring to the sampling points of networks.This paper presents the application of two empirical models for simulating andforecasting turbidity within drinking water distribution systems. The first is a multivariatelinear regressive model with data analysis tool SAS;the second is an artificial neuralnetwork model with numerical simulation tool MATLAB. The development of bothmodels is founded on representative data from the two drinking water systems mentionedabove.The results demonstrate both models have a satisfactory effect, and can be used tothe prediction of practical water supply system. In evaluating all the given data,simulation results show a similar performance for both of the models. However, forspecific treatment conditions of very high or very low turbidity, the artificial neuralnetwork model gives better prediction than the multivariate regression model.Representative data is the foundation of water quality empirical model so thatstrategies are designed to identify more representative data for future, based on theexisting state of the demonstrate areas.