基于多幅连续相关法的超声检测信号的缺陷识别技术  

Defect Recognition of Ultrasonic Testing Signals Based on Multi-Frame Correlation

作　　者：任明照[1] 黄永巍[1] 高东海[1] 

机构地区：[1]中国铁道科学研究院金属及化学研究所无损检测事业部,北京100081

出　　处：《无损检测》2012年第4期38-41,共4页Nondestructive Testing

摘　　要：相关法是时域中描述信号相似性特征的一种常用分析方法。文章基于数字相关法,针对超声检测中的大量检测回波信号,提出了一种多幅连续相关的处理方法,以实现超声检测信号的缺陷追踪与识别,并在实际检测试验中采用该方法对A型扫查的大量超声自动化检测信号进行处理,不仅有效地排除了随机干扰,同时保留了缺陷信号的完整信息,缺陷识别稳定、快速,识别准确率可达90%以上。Correlation is a useful method to describe the similarity between signals in time domain. Based on digital correlation, a new method of multi-frame correlation is introduced to process large amount of inspection signals and complete defect recognition in the experiment of ultrasonic testing, and in this process, the multi frame correlation method not only eliminates the random disturbing signals, but also acquires a high correct rate beyond 90% of defect recognition.

关 键 词：超声检测 相关法 缺陷识别 

分 类 号：TG115.28[金属学及工艺—物理冶金]