基于SVM的核动力管道腐蚀状态评估方法研究  

Research on Corrosion State Evaluation Technology of Nuclear Power Pipelines Based on Support Vector Machine

作　　者：夏志敏[1] 张大发[1] 陈永红[1] 张耀[1] 

机构地区：[1]海军工程大学核能科学与工程系,武汉430033

出　　处：《武汉理工大学学报（交通科学与工程版）》2010年第5期1014-1017,共4页Journal of Wuhan University of Technology（Transportation Science & Engineering）

基　　金：海军装备部科研项目资助

摘　　要：支持向量机(SVM)是根据统计学理论提出的一种新的通用机器学习方法,在解决小样本分类评估等工程实际问题中具有独特的优势.文中根据腐蚀状态与腐蚀影响因素之间复杂的映射关系,通过采用不同的核函数建立支持向量机分类评估模型对核动力某管道的腐蚀状态进行了评估.计算结果表明,采用RBF核函数的计算速度较快,结果较准确,运用支持向量机方法对核动力管道进行腐蚀评估在样本有限的情况下具有明显优势.Support Vector Machine（SVM）is a new kind of general machine learning methods based on statistical learning theory,with the unique advantage of solving the classification and evaluation in engineering practical problems as the case of limited examples.Aimed at the complex reflection relationship between corrosion state and corrosion influence factors,the corrosion state evaluation of certain pipeline was introduced through establishing the classification and evaluation model based on SVM with different kernel function.The study and analysis of the calculating result shows that the SVM method has an obvious advantage on the classification and evaluation of nuclear power pipelines at the case of limited examples,and with RBF kernel function it has a faster calculating speed and a better result.

关 键 词：支持向量机 核动力管道 腐蚀 状态评估 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]