摘要
针对珠江水质预测中的大量不确定和模糊因素,提出了一种基于属性重要性的加权支持向量机水质预测模型.首先通过粗糙集理论对原有的评价指标体系进行约简,由原来的8个预测指标约简为7个指标,被约去的属性正是网站公布数据中缺失的属性;同时计算出各属性的重要性,对重要的指标赋予较大的权重,构造基于属性重要性的加权支持向量机,这不同于以前的针对样本作用不同而构造的加权支持向量机.本文以珠江流域重点断面水质预测为例,对近2年数据进行分析,结果显示了该模型的有效性.
In view of many uncertainty and fuzzy factors in water quality prediction,a new weighted support vector machine water quality prediction model based on the importance of attributes is proposed in this paper.By attribute reduction,the prediction indexes are reduced to seven indexes from the original eight indexes,moreover,the reduced attribute is also the missed attribute reported in the official website;the weight coefficient of each index is calculated by attribute importance,the larger weight coefficient is proposed to give the important index,then we can construct the weighted support vector machine,which is different from the sample-based weighted support vector machine.Taking Pearl River water quality prediction for example,we analyze its nearly two years data,and the prediction result shows the efficiency and feasibility of the proposed model.
出处
《烟台大学学报（自然科学与工程版）》
CAS
北大核心
2011年第1期65-69,共5页
Journal of Yantai University(Natural Science and Engineering Edition)
基金
国家自然基金资助项目(10771213)
中国农业大学博士科研基金资助项目(2007038)
中央高校科研业务专项基金资助项目(2009-2-05)
关键词
评价指标
水质预测
粗糙集理论
属性重要性
加权支持向量机
evaluation index
water quality prediction
rough set theory
importance of attributes
weighted support vector machine
