基于分段线性表示k最近邻的水质预测方法  

Water quality prediction based on piecewise linear representation k nearest neighbor

作　　者：王保良[1] 范昊[1] 冀海峰[1] 黄志尧[1] 李海青[1] 

机构地区：[1]浙江大学控制科学与工程学院,工业控制技术国家重点实验室,杭州310027

出　　处：《环境工程学报》2016年第2期1005-1009,共5页Chinese Journal of Environmental Engineering

基　　金：国家“水体污染控制与治理”科技重大专项(2008ZX07420-004)

摘　　要：随着自动监测、网络通讯等技术的迅速发展,水质数据采集方式从原来的人工采集发展到现在的自动采集,技术上得到很大的进步,同时获得的水质数据也急剧增加。因此面对大量的水质数据,迫切需要一种能够处理大规模水质数据的预测方法。针对这一问题,基于k最近邻算法和分段线性表示算法,提出了分段线性表示k最近邻算法用于水质预测。为了验证所提出算法的有效性,利用该算法对2个水库进行水质浑浊度预测实验。实验结果表明,分段线性表示k最近邻算法处理大规模水质数据时可以有效减少计算量和运行时间,且预测效果令人满意。With the rapid development of automatic monitoring and network communication technique,etc.,the method of water quality data acquisition has been improved from manual acquisition to automatic instrument acquisition. That has made great progress in technique,and the obtained water quality data have been increasing rapidly. Therefore,a new prediction method should be proposed to deal with the large amount of water quality data. Based on piecewise linear representation algorithm and k nearest neighbor（ k NN） algorithm,the piecewise linear representation k nearest neighbor（ PLR-k NN） algorithm is proposed to implement the water quality prediction. The algorithm is verified by applying it for predicting the water turbidity in two reservoirs respectively. Experimental results show that the PLR-k NN algorithm can effectively reduce the computational burden and execution time,and the prediction accuracy is satisfactory.

关 键 词：水质预测 分段线性表示kNN算法 运行时间 

分 类 号：X824[环境科学与工程—环境工程]