基于K均值聚类的海洋平台T型管节点损伤声发射信号的模式识别  

Pattern Recognition of AE Signals of Offshore Platform T-tube Damage Based on K-means Clustering

作　　者：蒋鹏[1] 张璐莹[1] 李英年[1] 李伟[1] 

机构地区：[1]东北石油大学,大庆163318

出　　处：《无损检测》2015年第8期48-50,56,共4页Nondestructive Testing

基　　金：国家科技支撑计划资助项目(2012BAH28F03)

摘　　要：对海洋石油平台T型管节点损伤进行了声发射监测,并对损伤检测过程的声发射信号进行K均值聚类分析,以声发射参量的相关性为依据选取损伤特征参数,实现了对T型管节点不同损伤阶段信号的有效识别,为T型管节点的损伤监测和不同损伤阶段的模式识别提供了依据。An experiment was performed on the offshore platform T-tube node for the purpose of understanding its damage mechanism.Characteristics of acoustic emission signal of the offshore platform T-tube node at different damage stages were obtained by using the K-means clustering analysis.The damage characteristic parameters of the acoustic emission signal were first selected based on the correlation analysis.The proposed method could thus effectively realize the recognition of T-type pipe node at different damage stages.The study should provide reference for damage monitoring of T-type pipe node and the identification of its different damage stages.

关 键 词：T型管 声发射 K均值聚类 模式识别 

分 类 号：TG115.28[金属学及工艺—物理冶金]