基于聚类分析的罐底声发射检测信号融合方法

摘要

In acoustic emission (AE ) testing for a tank bottom, it is needed to process signals acquired from multi sensors mounted on the outside surface of the tank wall. During the signal processing procedure, the signals coming from the same AE source were grouped as an AE event. However, due to influence of noise signals, the traditional AE sources determining method often got wrong AE events. In order to avoid getting wrong AE events combined with the signals coming from different AE sources, a signal fusion method based on clustering analysis was proposed. The method was based on the classification of AE signals recorded in each event definition time. Each class containing three or more signals was determined as a target event. An experiment showed that the proposed method has strong ability to resist noise interference and low probability of misjudging AE events; therefore, it can raise the accuracy of tank bottom corrosion evaluation.%采用声发射技术进行罐底腐蚀与泄漏检测过程中,需要对多个传感器的检测到的声发射信号进行融合处理,将属于同一声发射源的声发射信号判定为一个声发射事件.但是在现场检测过程中,由于噪声的存在,使得在声发射信号融合处理时容易对声发射源产生误判.针对该问题,提出了一种基于聚类分析的罐底声发射信号融合方法,其基本原理是先根据事件定义时间进行初始声发射事件判定,然后采用聚类分析方法对初始声发射事件中的信号进行分类,将每一类信号分别判定为一个声发射源.现场实验表明采用该方法抗噪声干扰能力强、误判概率低,能准确反应罐底腐蚀的实际情况.