基于PCA-SVM的高含硫油气混输管道腐蚀预测  

Corrosion prediction of high sulfur gas-oil mixed transmission pipelines based on PCA-SVM

作　　者：骆正山[1] 毕傲睿 王小完[1] 

机构地区：[1]西安建筑科技大学管理学院

出　　处：《中国安全科学学报》2016年第2期85-90,共6页China Safety Science Journal

基　　金：国家自然科学基金资助(61271278);陕西省重点学科建设专项资金资助(E08001);陕西省教育厅自然科学基金资助

摘　　要：为预测高含硫油气混输管道腐蚀情况,分析了导致腐蚀的原因,归纳影响腐蚀的因素。采用主成分分析法(PCA)对各种因素进行优选,摒弃相关联但贡献率较低的因素。将贡献率较大的因素作为支持向量机(SVM)的输入变量,以腐蚀率作为目标输出函数,建立管道腐蚀预测模型,并进行管道腐蚀率预测。以某公司川中地区8条运营管线为例,验证SVM方法预测管道腐蚀率的效果。结果表明,PCA-SVM方法的预测数据平均相对误差较小、吻合度高,预测结果符合实际情况。In order to predict and analyze corrosion situations of high sulfur gas pipelines,analyzed causes of corrosion were analyzed,and factors affecting corrosion were identified. All the factors were filtered by using the PCA method. Factors making small contributions were discarded. Factors making big contributions were taken which higher contribution as the input variables of SVM,the corrosion rate as the target output function,and a model was built for predicting corrosion of pipelines. The model was used to predict pipeline corrosion rates. Eight operating pipelines in the central region of Sichuan province were taken as an example of objects of prediction by the SVM method. The results show that the data predicted by PCASVM method have lower average relative error and conform with the practical situation.

关 键 词：高含硫混输管道 腐蚀因素 腐蚀率预测 主成分分析(PCA) 支持向量机(SVM) 

分 类 号：X937[环境科学与工程—安全科学]