摘要
声发射信号识别目前已经成为检测大型储罐底板缺陷状况的普遍方法之一。针对现有声发射检测技术对原始信号的处理与分析方法的不足,对储罐底板腐蚀程度分类识别方法的不明确,基于小波包分解,对声发射传感器接收到的原始信号进行特征提取,得出了信号的特征向量,通过训练相关向量机,结合相关向量机模型对待测设备进行腐蚀类型归类识别,得出分析结论。应用该方法对大连某企业的储油罐进行实测数据检验,分析结论与储油罐实际状况相同,验证了该方法具有科学性与可行性,能够诊断大型储罐底板腐蚀状况的准确性和可靠性。
Acoustic emission signal recognition is one of the common methods to detect the defects of large storage tank floor. In order to overcome the deficiency of the existing acoustic emission detection technology to the original signal processing and analysis method, and to identify the corrosion degree of the tank bottom plate, based on the wavelet packet decomposition, the original signal received by the acoustic emission sensor is ex- tracted, and the characteristic vector of the signal is obtained. Through the training of the relevance vector ma- chine（RVM）, combined with the RVM, the types of corrosion of the tested equipment are identified, and the conclusion of the analysis is drawn. This method is used to test the measured data of oil tank of an enterprise in Dalian, and the same analysis conclusion with the actual situation of oil tank is obtained. It is proved that the method is scientific and feasible, and it can diagnose the accuracy and reliability of the corrosion of the bottom plate of large storage tank.
作者
许宇峰
孙铁
周长茂
张素香
XU Yu-feng SUN Tie ZHOU Chang-mao ZHANG Su-xiang(School of Mechanical Engineering, Liaoning Shihua University, Fushun 113001, China)
出处
《测控技术》
CSCD
2017年第7期17-21,共5页
Measurement & Control Technology
基金
中国石油化工股份有限公司科技攻关项目(313084)
关键词
声发射信号识别
小波包分解
相关向量机
acoustic emission signal recognition
wavelet packet decomposition
relevance vector machine
