摘要
目的应用支持向量机预测水中化学耗氧量(COD)浓度。方法基于研究地区的水质监测数据,运用支持向量机算法构建水质预测模型,并对某水厂水源水中COD月均浓度进行了预测。结果建模以后预测结果显示,训练集的均方误差MSE=0.039 4,相关系数R=0.805 4,平均相对误差MRE=4.22%;两组测试集样本的相对误差分别为8.89%和0.98%。结论该模型的预测性能良好,简便易行,可以为水质预测提供新思路。
Objective To predict COD content in source water by applying support vector machines（SVM）. Methods An SVM model was established to predict water quality based on monitoring data, and it was used to predict monthly mean COD contents from source water of a water supply facility. Results The study outputs had shown that the mean squared error（MSE）,correlation coefficient（R） and mean relative error（MRE） were 0.0394, 0.8054, 4.22% in the train set respectively;In the two samples of test set, the MREs were 8.89% and 0.98%, respectively. Conclusion This model shows good performance in this case. This method provides reference for water quality prediction as it is simple and available.
出处
《环境与健康杂志》
CAS
北大核心
2016年第6期544-547,共4页
Journal of Environment and Health
基金
原卫生部卫生公益性行业科研专项(201302004)
关键词
支持向量机
水质
化学耗氧量
预测
Support vector machines
Water quality
Chemical oxygen demand
Prediction
作者简介
孙伯寅（1974-），男，助理研究员，从事水与环境卫生研究。
通讯作者：张荣，E-mail：rzhang@crwstc．org
