基于加权支持向量机的焊缝RT图像中缺陷分类研究  

Study to Classification of the Defect Presented on the RT Photo of the Welding Seam on Basis of the Weighted Support Vector Machine

作　　者：刘元祥[1,2] 张晓光[1] 匡颖芝[1] 张长明[1] 李国庆[1,3] 

机构地区：[1]中国矿业大学机电工程学院,江苏徐州221008 [2]汾西矿业集团公司柳湾煤矿 [3]平顶山煤业集团三环有限公司

出　　处：《矿山机械》2006年第8期104-106,共3页Mining & Processing Equipment

基　　金：江苏省博士后科研基金资助课题(2004035);中国矿业大学科技基金资助课题(2005B005)。

摘　　要：分析了支持向量机的分类原理,指出在各类别样本数目相差较悬殊时,SVM不能获得良好的分类能力。针对焊接缺陷分类,提出了加权SVM(WSVM)算法。测试结果表明,该算法在焊缝RT图像中缺陷的分类识别中,能提高小类别缺陷的的检测精度,具有较高的理论和应用价值。The paper analyses the classifying principle of the support vector machine(SVM),indicates that the SVM could not bear the excellent classifying capacity while the number of each kind of the sample differ greatly.According to the defect classification of the welding seam,the paper puts forward the weighted SVM algorithm.The testing result shows that the algorithm could increase the detection accuracy of the small kind defect on identifying the classification of the defect presented on the RT photo of the welding seam,which has higher theoretical and applicable value.

关 键 词：缺陷定性 焊缝缺陷 焊缝图像 支持向量机 分类研究 模糊神经网络 RT 加权 无损检测 焊接结构 

分 类 号：TG441.7[金属学及工艺—焊接] TN247[电子电信—物理电子学]