摘要
饮用水源健康风险非确定性的定量化对水安全的准确评估十分重要。为减小非确定性对健康风险的影响,基于美国环境保护署推荐的健康风险评价模型,将人工神经网络和蒙特卡洛分析方法进行耦合,构建了ANN-MC健康风险评估模型,并对四川盆地西南缘周山地典型区的农村饮用水源进行健康风险评价。结果表明:在2010年、2011年、2012年,样本非致癌风险高于可接受水平"1.0"的仅为0.31%、0.04%和0.04%,而在0.1≤HQ<1.0之间的样本数分别为95.00%、78.75%和66.80%,说明研究区域内大部分居民可能通过饮水途径造成健康危害;该区农村饮用水源中需要重点控制的污染物为氟化物、硝酸盐、铁、锰和铜。ANN-MC模型预测健康风险是一种有效评估饮用水源健康风险的方法,可为水安全管理提供更加可靠的科学依据。
The quantification of uncertainty analysis in health risk associated with drinking water sources is very important to accurately estimate water safety. In order to reduce the impact of the uncertainty on the results of health risk assessment, in the paper, based on the model of health risk assessment recommended by the United States Environmental Protection Agency （USEPA） , and combined the methods of Artificial Neural Network （ANN） and Monte Carlo （MC） , the model of ANN-MC health risk assessment was constructed, then the model was used to estimate the health risk associated with the rural drinking water sources of typical mountain area in the southwest edge of Sichuan Basin. Results show that in 2010, 2011 and 2012, the sample proportions that the non-carcinogenic risk of the samples is higher than the acceptable level 1.0 are only 0.31% , 0.04% and 0.04% respectively, and the sample proportions that the level of the non-carcinogenic risk of the samples is 0.1 ~〈 HQ 〈 0.1 are 95 % , 78.75% and 66.8% respectively, it means that most of residents＇ health may be threatened by drinking water; The main pollutants in rural drinking water sources of study area are fluoride, nitrate, iron, manganese and copper. The ANN-MC method which has been used to forecast health risk is an effective method to assess the health risks of drinking water sources, and the assessment results can provide more reliable scientific basis for the management of water safety.
出处
《华北水利水电大学学报（自然科学版）》
2017年第4期25-32,共8页
Journal of North China University of Water Resources and Electric Power：Natural Science Edition
基金
国际科技合作项目(2012DFG91520-3)
四川省教育厅"农村水安全"四川省高等学校工程研究中心项目(035Z2020)
关键词
饮用水源
健康风险
蒙特卡洛
人工神经网络
非致癌风险
drinking water sources
health risk
Monte Carlo
neural network
non-carcinogenic risk
