Research On Coagulation And Dosing Control System Of Water Plant Based On BP Neural Network

In modern life and production, with the fast development of economy, the requirement of safety drinking water for urban residents is increasing day by day, the burden of urban water supply department is more and more heavy, how can we ensure good water quality and reduce the product cost in the process of water supply has become a hot topic in water treatment industry.Coagulating sedimentation is the most important link in water treatment progress, coagulation effect is good or bad not only directly related to the subsequent treatment progress and also has great influence on the water quality of the whole system. As the result, the key task of water treatment progress is how can we have a good coagulation effect. When the coagulation agent is set, coagulation facility hydraulic conditions are set, still can ensure the quality of water and control the dosing amount of coagulation according to the changing quality of raw water has become the key indicator in judging a waterworks production.This topic based on a system of coagulation dosage in water plant in Kunming city as the research background, for the characteristics of nonlinear, large delay, much interference and multiple variable in waterworks system, after studying traditional coagulant dosage control scheme, combined with the theory and control characteristics of neural network, try to establish a coagulant dosage model based on BP neural network, and have a preliminary simulation research about coagulant dosage.This paper introduces the common methods of coagulant dosage at home and aboard at first, and have a comparative analysis about their characteristics one by one, then introduces the algorithm, structure features and modeling process of BP neural network, with a Levenberg-Marquardt algorithm improvement in the traditional BP neural network algorithm, on the basis of mass water production data, built a BP neural network with raw water turbidity, flow rate as input factors, dosing amount of coagulant factors as the output, and has carried out the simulation and prediction analysis. Based on the BP neural network controller, combined with the mathematical model of controlled object, good set PID controller, realized the simulation of the feed forward and feedback controller in Simulink, a good control effect can be seen in the simulation effect picture.