基于人工神经网络构建湖南省企业重金属废水预测模型研究  

Research on Constructing a Heavy Metal Wastewater Prediction Model for Enterprises in Hunan Province Based on Artificial Neural Networks

作　　者：曾钰 肖曲 陶佳 高雯媛 张翔宇 ZENG Yu;XIAO Qu;TAO Jia;GAO Wenyuan;ZHANG Xiangyu(Hunan Environmental Monitoring Center Station,Changsha,Hunan Province,410014 China;Key Laboratory of Heavy Metal Pollution Monitoring for National Environmental Protection,Changsha,Hunan Province,410014 China;Hunan Ecological Environment Affairs Center,Changsha,Hunan Province,410014 China;School of Traffic and Transportation Engineering,Central South University,Changsha,Hunan Province,410075 China)

机构地区：[1]湖南省环境监测中心站,湖南长沙410014 [2]国家环境保护重金属污染监测重点实验室,湖南长沙410014 [3]湖南省生态环境事务中心,湖南长沙410014 [4]中南大学交通运输工程学院,湖南长沙410075

出　　处：《科技资讯》2024年第24期202-204,共3页Science & Technology Information

摘　　要：基于径向基函数神经网络(Radial Basis Function Neural Network,RBFNN)构建了一种环境废水质量预测模型,根据湖南省2021—2022年监测的pH值、化学需氧量(Chemical Oxygen Demand,COD)和氨氮浓度数据,预测了2023年各月废水排放的环境质量。模型采用RBFNN单隐层结构,利用Softmax激活函数和梯度下降优化算法对废水数据进行了建模和预测。实验结果表明,pH值在7.2~7.4之间,氨氮浓度稳定在国家一级排放标准以下,COD浓度虽有波动但总体符合国家标准。与传统模型相比,RBFNN能够更好地捕捉数据中的非线性特征,提高了预测精度,展示了RBFNN在环境科学中的应用潜力,并为废水质量预测提供了有效的技术路径。A prediction model for environmental wastewater quality was constructed based on Radial Basis Func-tion Neural Network(RBFNN).and it predicted the environmental quality of wastewater discharge in each month of 2023 using pH value,Chemical Oxygen Demand(COD),and ammonia nitrogen concentration data monitored in Hu'nan Province from 2021 to 2022.The model adopted RBFNN single hidden layer structure and used Soft-max activation function and gradient descent optimization algorithm to model and predict wastewater data.The ex-perimental results showed that the pH value was between 7.2 and 7.4,and the ammonia nitrogen concentration re-mained stable below the national first level emission standard.Although the COD concentration fluctuated,it over-all met the national standard.Compared with traditional models,RBFNN can better capture nonlinear features in data,improve prediction accuracy,demonstrate the potential application of RBFNN in environmental science and provide an effective technical pathway for predicting wastewater quality.

关 键 词：径向基神经网络 重金属污染 预测模型 废水排放 

分 类 号：X84[环境科学与工程—环境工程]