Prediction Research Of Industrial Circulating Cooling Water Based On BP Artificial Neural Network

In today’s circulation industry cooling water treatment technology, the way to deal withcooling water mainly adopts the manual dosing. Due to the uncertainty of human experience,it will cause the waste of water resources and the increase of economic cost, in today’senvironment, the quantity demanded of industrial water has become increasingly tense,predicting the water quality change trend of cooling water and gives a scientific watertreatment plan is particularly important. In this paper, based on the large amounts of datawhich is accumulated by TianJin petrochemical company Industrial Water InformationManagement System, direct at difficulties of industrial cooling water with high complexity,uncertainty and nonlinear measurement, artificial neural network is used to establish themodel of water quality of circulating cooling water. It is one of the key points in this paperthat predicting the corrosion rate and scaling rate of equipment which are caused by waterquality change, it has important theoretical significance.In this paper, through the research about the common problems and the industrialcirculating cooling water basic structure, analyses the control method for sediment, metalcorrosion, microbial slime and enrichment ratio. Then, based on the water quality data ofTianJin petrochemical company, use principal component analysis to carry out the waterquality data preprocessing, then use the BP neural network to establish the forecast model ofwater quality, and the water quality model is simulated and tested on MATLAB platform.Aiming at the disadvantage existed in the work of the application that the choice of networkrandom initial weights and threshold cannot be based on an effective selection mechanism,the difficulty in determining the overall excellent initial network parameters and the speed isslow convergence, however, the genetic algorithm has great advantage in global search, souse genetic algorithm to optimize the weights and threshold of BP neural network, replacesthe random initial network parameters values in the traditional method to make neuralnetwork modeling operations within a relatively optimal search collection. Lastly, running thealgorithm on the basis of the optimization, achieving the goal through learning that make theprediction results of water quality and the actual situation more fitting.The new method of intelligent control algorithm was come up in this paper to predictwater quality of the industrial circulating cooling water, the prediction accuracy is higherunder the condition of the definition of water quality. Finally using the method of C#andMATLAB mixed programming to develop the circulating cooling water intelligence auxiliaryanalysis platform, through the design of the user login module, water quality of pretreatmentmodule, water quality prediction module, experts suggested module, auxiliary informationquery module and help module, it has realized the basic function of the water qualityprediction and given advice of experts for different forecast results, at the same time, hadgood guidance to practical work.