网络安全与数据治理2025，Vol.44Issue(5)：29-34,6.DOI:10.19358/j.issn.2097-1788.2025.05.005
融合卷积神经网络的混凝投药模型研究
Research on a coagulation dosing model incorporating convolutional neural networks
摘要
Abstract
This study focuses on a water treatment plant in a southeastern Chinese city with a population of one million.To ad-dress the inefficiency of traditional water quality monitoring and the difficulty in pre-determining coagulant dosage,we proposed a prediction model based on convolutional neural networks(CNN).After enhancing data quality through preprocessing,key fea-tures were selected using the information gain ratio.A CNN architecture incorporating 1D convolutional layers,pooling layers,and fully connected layers was developed,with ReLU activation functions optimizing feature representation.Experimental results demonstrated superior performance of the model,achieving a root mean square error(RMSE)of 68.550,mean absolute error(MAE)of 50.709,and goodness-of-fit of 0.926.关键词
水质监测/混凝剂预测/卷积神经网络Key words
water quality monitoring/coagulant prediction/convolutional neural network分类
计算机与自动化引用本文复制引用
李泽楷,章杰..融合卷积神经网络的混凝投药模型研究[J].网络安全与数据治理,2025,44(5):29-34,6.基金项目
福建省级科技创新重点项目(2022G02011) （2022G02011）
