摘要
研究了304控氮不锈钢试样在酸性氯化钠溶液中慢应变速率拉伸过程的声发射特征。采用基于自组织映射神经网络和K-均值聚类算法对长时间慢拉伸实验的声发射信号进行聚类分析,通过分析各类信号的持续时间、上升时间、振铃、能量、幅值、波形、频带能量等特征,从中找出了裂纹信号。将分类后的信号作为样本训练神经网络,对短时间慢拉伸实验检测到的声发射信号进行识别,找出了应力腐蚀初期的裂纹萌生信号,且与长时间慢拉伸实验检测到的声发射信号特征一致。研究结果表明,304控氮不锈钢应力腐蚀过程中的声发射测试结果与电化学噪声测试结果一致;304控氮不锈钢在酸性氯化钠溶液中的应力腐蚀过程主要会产生包括裂纹在内的3类声发射信号,通过聚类分析方法可以将3类声发射信号区分出来,找出裂纹信号。
The acoustic emission(AE)signals characteristic of nitrogen-added 304NG stainless steel stress corrosion process was studied.Two experiments were performed,one was long term slow strain rate test(SSRT),and the other was short term slow strain rate test.The solution was made of 0.5 mol·L-1 NaCl and 1.5 mol·L-1 H2SO4.By using combined self-organizing maps and K-means cluster algorithm to classify the AE signals in long term SSRT,in which the duration,counts,energy,amplitude,waveform and frequency band energy were analyzed as the AE signals' characteristics,and the crack AE signals were found.The classified signals were trained by using the artificial neural network,so as to identify the different types AE signals in short term SSRT.The result showed that AE test for 304NG stainless steel stress corrosion process would be confirmed by electrochemical noise test.And there were mainly three types of AE signals during 304NG stainless steel stress corrosion process,including crack,pitting and bubbles,which could be classified by using cluster analysis and be provided as template to build artificial neural network to identify the AE signals generated in the same corrosion environment.
出处
《化工学报》
EI
CAS
CSCD
北大核心
2011年第4期1027-1033,共7页
CIESC Journal
基金
国家重点基础研究发展计划项目(2006CB605004)~~