Prediction of effluent water quality based on 5-parameter GA-BP model——A case study of certain water plant in Ningxia
To solve the problems of large time delay and low precision in the sampling and detection of key water quality parameters in drinking water treatment,a turbidity prediction model based on genetic algorithm optimized BP neural network(GA-BP)is proposed.Using the measured turbidity and related water quality data of a waterworks in Yinchuan City from 2019 to 2021,the input indicators that affect the turbidity are screened using gray correlation analysis,and the sample data is clustered into three different groups with different characteristics using Q-type clustering analysis.A machine learning model based on GA-BP neural network is constructed to predict the turbidity,and the results are compared with those of traditional BP and unclassified prediction models.The results show that compared with unclassified prediction,the error evaluation indexes of the prediction model after Q-type clustering analysis are improved by 2.9%and 22%in terms of R2 and RMSE,respectively;compared with the traditional BP neural network,the error evaluation indexes of the prediction model optimized by genetic algorithm are improved by 2.4%and 12%in terms of R2 and RMSE,respectively.The research shows that both Q-type clustering analysis and genetic algorithm can improve the generalization ability of BP neural network prediction model and reduce errors.
BP neural networkgenetic algorithmcluster analysisgrey relation analysisturbidity prediction
国家科技期刊平台
NSTL
万方数据
