Research And Application Of Water Quality Prediction Model In City Water Supply Networks

For the purpose of researching water quality variation in water supply networksand improving the management of water quality, experiments and researches are madeto evaluate water quality of the selected network. Different residual chlorine decaymodels are formed, compared and analyzed. The Water quality information systembased on Geographic Information System is worked out.Firstly, the distribution network of a northern university is chosen as the studyobject. After more-than-one-month continual monitor, water quality status of thisnetwork is investigated through analyzing key parameters and their influencing factors.Understanding of residual chlorine decay rules is pushed forward. Suggestions ofimproving water quality status are provided.Secondly, hydraulic resident time among checkpoints is achieved by hydraulicanalysis, which is calculated with water consumption of the university and basic dataof the network. According to the calculated resident time, a second-roundmeasurement of the network is put into effect. Based on the former research, multiplelinear regression model and artificial neural network model of residual chlorine forcheckpoint are established for the first time. And support vector machine algorithm isinnovative introduced in water quality model of the distribution network. Comparedwith the first-order kinetics equation, SVM model works reasonably well, whichmake it possible to form water quality model in distribution system by small sample.Finally, water quality information system in distribution network is establishedthrough combining water quality model with Geographic Information System. Bycontrolling and analyzing water quality in distribution system visually, this systemmakes up the shortage of traditional water quality management. The work is availableexploration to improving water quality management of distribution system.