Study On The Prediction And Visualization System Of Sewage Quality Based On Neural Network

At present,the problem of water environment is becoming more and more serious in our country,and sewage treatment has become an important measure to protect water resources.Because the sewage treatment system has high uncertainty and strong nonlinearity,it is a difficult problem to measure the accurate monitoring of the key water quality parameters in the effluent.For the measurement of some key water quality parameters of the effluent,the existing measurement methods or instruments have many limitations and inconveniences,such as the detection week Long term,expensive instrument,cumbersome operation and other issues.The advantages of soft sensing technology are good timeliness,high precision and low cost.When it is applied to sewage treatment system,it can achieve an effective method to detect the key water quality parameters of effluent with high precision.Because the neural network modeling method has excellent approximation ability,it has been widely used in the soft sensor modeling of effluent quality parameters of sewage treatment.In this paper,COD and ammonia nitrogen of effluent are used as prediction indexes,and the neural network of gate control unit(GRU)is used to establish the prediction model of effluent water quality,and the visualization platform of water quality prediction is designed and realized,in order to apply it to the actual wastewater treatment production environment.The main contents of this paper are as follows:(1)According to the predicted COD and ammonia nitrogen in the effluent,the reasonable auxiliary variables were selected.First of all,the mechanism of the treatment process and the reaction process of the main parameters of the sewage treatment plant is understood and studied;then,by consulting the literature and analyzing the mechanism of the process,the variables with high correlation with the main parameters are initially selected as auxiliary variables;finally,PCA is used to eliminate the correlation between variables,eliminate the noise and redundancy in the data,and simplify the original complexity of process data again.The final parametersof auxiliary variables are obtained.(2)A water quality prediction model based on GRU neural network is established.Taking the actual historical sewage data of a sewage treatment plant in Lujiang County of Hefei as the experimental data and the sewage index closely related to the water quality index to be tested as the auxiliary variable,a prediction model of GRU neural network with COD and ammonia nitrogen as the main variables of the effluent is established,and the model is improved by using the adaptive learning rate method.Finally,the predicted results of the model are compared with the experimental data Compared with the real value,the error is within the acceptable range,which proves the effectiveness of GRU neural network training.(3)A SVM-GRU water quality prediction model based on SVM and GRU neural network is established.Aiming at the problem that the neural network model is easy to fall into local optimum and difficult to achieve global optimum in the training process,the support vector machine classification model is introduced to make the change of the neural network when learning the concentration of parameters to be measured in a certain range,so as to reduce the influence of local optimum on the global scope,and then achieve the purpose of improving the prediction performance of the model.In the SVM model training,the grid search method and cross validation method are used to optimize the model parameters.The final SVM-GRU joint prediction model has more accurate prediction accuracy,better model effect,network performance can meet the actual application needs,and can achieve accurate prediction of the effluent quality of the sewage treatment system.(4)Establish the prediction system of key wastewater quality parameters.In order to apply the above achievements to the actual sewage treatment of sewage treatment plants and enterprises,the key water quality prediction system is designed and implemented.First of all,through the research and analysis of the development needs and objectives of the sewage data analysis system,a detailed system framework is designed,and then the functional design and program development of the whole system are carried out.The system can be applied to the actual sewage treatmentscene after the future design optimization,providing a new scheme for the sewage treatment of urban sewage treatment plants and enterprises.