Prediction Of IC Reactor Wastewater Treatment System Based On BP Neural Network

Internal Circulation(IC)reactor as the third generation of anaerobic reactor have the advantages of high volume load,land saving,stable operation and convenient maintenance,as well as some of the recycling the energy from sewage,which will make up for the consumption of energy.Under the pressure of resources and energy,if the whole IC reactor can be accurately modeled,it can effectively control and adjust the treatment process.Due to the wastewater treatment process is a nonlinear,large lagging,multi-factor coupling process,which is difficult to establish precise model.In this paper,an artificial neural network(ANN)model has been established to forecast the IC reactor ’s running effect in advance.This paper predicts the effluent COD of an IC reactor in a papermaking wastewater treatment plant in Shenzhen.The prediction results are as follows:(1)The BP neural network prediction model of IC reactor in advance take water flow,COD,temperature and VFA at 9 am.as input factors,the effluent COD at 4 pm as the output.The average absolute error value is 0.0720 and the maximum relative error is 0.1860.It shows that the established BP neural network is able to capture the nonlinear mapping between the selected input factors and the output and the COD value in the IC reactor can be predicted in advance.It also indicates that the selected input factors are reasonable.(2)Due to the BPNN is influenced by the initial value,easily trapped in local small shortcomings,a genetic algorithm(GA)has been chosen to optimize the BP model.The average absolute error value is 0.0491,the maximum relative error is 0.1508.Compared BPNN model and GA-BP model,the prediction error within 5% is increased from 45% to 65%,and the prediction error within 10% is increased from 75% to 90%.which realize the effective prediction of effluent COD value in the IC reactor in advance and can provide guidance for the adjustment of the reactor and the setting of the subsequent process conditions.(3)The adaptability and reliability of the model for other IC reactors is verified.the results show that the GA-BPNN model for the treatment of papermaking wastewater of other operating conditions of the IC reactor has good prediction performance.