起重机钢梁疲劳特性声发射监测实验研究

起重机钢梁作为主要承力部位关系到起重机的安全运行,目前仍无有效手段对其疲劳特性进行在线实时监测。采用声发射技术对钢梁材料Q345疲劳特性进行实验研究,首先通过动态弯曲疲劳实验获得材料疲劳裂纹萌生、扩展和断裂全过程的声发射检测信号;然后提取和分析声发射信号中的特征参数,经过比较各参数历程图,发现在幅值和事件历程图中可以反映材料疲劳裂纹整体演变过程,而能量、计数、上升时间和持续时间这4个特征参数更能反映出裂纹变化的重要转折点。此外对各个阶段声发射信号产生的原因进行分析和总结,为下一步采用声发射技术对钢梁材料损伤定量及寿命预测的研究提供参考依据。     

基于VMD-PNN的砂轮钝化声发射检测

在磨削加工过程中,加工刀具即砂轮会发生钝化现象,砂轮表面磨损影响加工精度和工件质量,需要及时检测并修整。磨粒的塑性变形、破碎、断裂等会产生声发射信号,能够作为精确识别砂轮钝化状态的依据,且不易被噪声干扰,因此提出一种基于变分模态分解（Variational Mode Decomposition,VMD）和概率神经网络（Probabilistic Neural Network,PNN）的砂轮钝化声发射检测方法。VMD可以将原始信号分解为多个本征模态函数（Intrinsic Mode Function,IMF）分量,筛选其中峭度较大的分量重构即得到声发射信号。声发射检测的关键是特征参数的选取,在相关研究基础上本文提出了声发射包络能量占比作为一个重要的特征参数,并选取了共5种特征参数,构建出5维特征向量数据集,输入到PNN中进行训练,经过测试识别准确度达到94.5%。该方法建立了声发射信号特征参数与砂轮不同钝化状态的关系,能够对砂轮严重钝化状态给出准确预警,具有实际应用价值。文章比较了声发射信号不同特征参数用于识别砂轮钝化状态的准确度,对特征参数的选用具有参考意义。     

基于声发射及WST-CNN协同的滑动轴承润滑状态识别

为实现声发射信号对滑动轴承润滑状态变化进行灵敏表征,提出一种采用小波散射变换及卷积神经网络结合的滑动轴承润滑状态识别及故障诊断研究方法。以某310 MW汽轮发电组滑动轴承现场试验所得声发射信号为研究对象,将现有小波散射网络加入散射路径优化机制并进行参数优化,对滑动轴承声发射信号进行自动鲁棒特征提取,将最佳特征矩阵输入优化后的卷积神经网络进行润滑状态识别分类。结果表明,优化后的小波散射网络能够有效提取声发射信号特征,结合优化后的卷积神经网络对特征矩阵进行智能识别,对滑动轴承润滑状态识别率可达到95.28%,能够高效精确地对滑动轴承润滑状态进行诊断。     

以声发射信号的特征诊断敏化304不锈钢沿晶应力腐蚀

采用声发射检测技术研究了304不锈钢在0.5 mol/L Na2S2O3溶液中的沿晶应力腐蚀过程,根据声发射信号的平均频率及幅值等特征参数,按声发射源将其分为易感晶界的溶解、裂尖的塑性变形及韧带的机械断裂3个阶段,并采用参数关联法及时频法聚类分析了3个阶段的信号特征。试样现场断口形貌及信号特征参数分析结果表明,声发射技术能够用于监测沿晶应力腐蚀演化进程,辨别其微观发展,为其微观机制研究提供依据,实现应力腐蚀在线精确监测。     

新型木塑材料缺陷及损伤的声发射信号分析

综合新型木塑复合材料各类模式试样、源定位及信号的波形、常规参数、频谱、小波包最优树叶子节点能量谱等特征,对主损伤区附近的声发射事件,应用频谱分析和小波变换等信号处理手段提取特征参数,确定不同缺陷及损伤模式所对应的声发射特征信号,为日后进行神经网络模式识别奠定基础。由于新型木塑复合材料的声发射研究刚刚起步,对该材料的声发射特征还有待进一步的分析,常常需要借鉴其它复合材料的声发射检测结果,这势必会带来一定的局限性及适用性问题。对新型木塑复合材料的声发射参数的定量化还有待于大量实验数据的积累和归纳分析。     

基于声发射理论的阀门气体内漏量化检测研究

通过理论分析和实验研究的方法,建立阀门内漏过程中气体体积泄漏率与声发射信号特征参数均方根(AERMS)量化关系。利用研制的实验平台对阀门气体内漏进行检测实验,并探讨了泄漏率、阀门类型等参数对声发射信号特征参数均方根的影响。实验表明阀门发生气体泄漏时产生的声发射特征信号参数均方根能有效反应气体体积泄漏率,且声发射检测技术对阀门泄漏率检测误差在10%以内,因此可以利用声发射技术检测阀门是否内漏并估算其泄漏率。     

基于模糊神经网络与主成分分析的磨削表面粗糙度在线预测

基于声发射和振动信号提出了一种模糊神经网络和主成分分析的表面粗糙度预测方法,以提高磨削过程中工件表面粗糙度识别的准确性。首先,采集磨削程中声发射与振动信号,提取相关时域特征、频域特征和小波包特征参数,利用主成分分析对特征量进行降维优化;然后,构建表面粗糙度模糊神经网络预测模型,将信号特征量与表面粗糙度作为模糊神经网络的输入和输出;最后,对模型进行训练,并对表面粗糙度预测精度进行验证。实验结果表明:通过主成分分析(PCA)方法对声发射和振动信号特征量进行降维得到5个主成分,以此建立的模糊神经网络表面粗糙度预测模型的效果精度可达到91%以上,与局部线性嵌入和多维标度法降维方法相比,PCA方法降维后的特征所含信息更优,预测准确度更高。     

基于BP神经网络的金属拉深件裂纹在线监测

论文运用设计的三层BP神经网络对采集到的10个声发射参数进行特征提取。通过对比不同隐含层神经元个数的BP神经网络的训练误差与训练次数,确定当隐含层神经元个数为13个时,BP神经网络的逼近效果较好,产生的网络误差最小。然后利用计算各声发射参数对表征裂纹信号灵敏度的大小,逐步删除各个声发射参数,降低模式识别时输入信号的维数。最后确定相对到达时间、幅度、能率、上升计数、持续时间和平均信号电平六个声发射参数能够有效地识别金属拉深件裂纹。本研究对于金属拉深件裂纹的在线监测具有理论和实际意义。     

基于小波变换与统计分析的转子碰摩声发射特性研究

为解决转子碰摩损伤声发射信号分类及解释难题,应用PCI-2声发射系统和WS-ZHT1型多功能转子实验台组成转子碰摩声发射检测系统,采集转子局部碰摩声发射信号,通过理论分析声发射信号特征和小波基函数性质,dB8小波适合提取声发射信号特征;碰摩产生大量声发射信号,大量声发射信号的统计特性蕴涵较多碰摩信息。对不同转速条件下不同检测位置碰摩声发射信号的统计分析表明,声发射信号的功率谱密度集中在100～400 kHz。声发射信号平均幅值、平均能量可作为区分转子碰摩程度特征参数;功率谱主频可作为区分声发射相对位置特征参数,结论与碰摩类型无关。     

海洋平台油气管道疲劳裂纹AE信号特征提取及识别研究

为了将声发射(AE)技术实际应用到监测海洋平台油气管道疲劳裂纹中,需要解决管道振动干扰以及疲劳裂纹AE信号有效特征提取的问题,而问题的关键在于对管道结构疲劳裂纹AE信号特征提取及识别算法的研究。在已有研究的基础上,提出了一种基于经验模态分解(EMD)为特征提取的疲劳裂纹识别方法,将管道振动干扰问题和疲劳裂纹AE信号有效特征提取问题联系在一起,对特征元素进行优化并剔除无效噪声干扰信息,通过概率神经网络(PNN)对疲劳裂纹信号进行识别。试验结果表明,PNN结合基于EMD为特征提取的疲劳裂纹识别法能够取得良好的效果,为声发射技术监测海洋平台油气管道疲劳裂纹提供了试验和理论依据。     

Detectability of Crack Lengths from Acoustic Emissions Using Physics of Wave Propagation in Plate Structures

This paper presents a study to understand the physical nature of fatigue crack growth as an acoustic emission source and detectability of the crack length form the recorded acoustic emission signal in plate structures. For most of the thin walled engineering structures, the acoustic emission detection through sensor network has been well established. However, the majority of the research is focused on prediction of the acoustic emission due to fatigue crack growth using stochastic methods. Where, stochastic models are used to predict the criticality of the damage. The scope of this research is to use predictive simulation method for acoustic emission signals and extract the damage related information from acoustic emission signals based on physics of material. This approach is in contrast with the traditional approach involving statistics of acoustic emissions and their relation with damage criticality. In this article, first, we present our approach to understand fatigue crack growth as source of acoustic emission using physics of guided wave propagation in FEM. Then, using this physical understanding, we present our investigation on detectability of crack lengths directly from crack-generated acoustic emission signals. Finally, we present our method to extract fatigue crack length information from acoustic emission signals recorded during fatigue crack growth.     

Acoustic emission location on aluminum alloy structure by using FBG sensors and PSO method

Acoustic emission location is important for finding the structural crack and ensuring the structural safety. In this paper, an acoustic emission location method by using fiber Bragg grating (FBG) sensors and particle swarm optimization (PSO) algorithm were investigated. Four FBG sensors were used to form a sensing network to detect the acoustic emission signals. According to the signals, the quadrilateral array location equations were established. By analyzing the acoustic emission signal propagation characteristics, the solution of location equations was converted to an optimization problem. Thus, acoustic emission location can be achieved by using an improved PSO algorithm, which was realized by using the information fusion of multiple standards PSO, to solve the optimization problem. Finally, acoustic emission location system was established and verified on an aluminum alloy plate. The experimental results showed that the average location error was 0.010 m. This paper provided a reliable method for aluminum alloy structural acoustic emission location.     

Prediction of low-cycle fatigue-life by acoustic emission—1: 2024-T3 aluminum alloy

Low-cycle fatigue tests were conducted by tension-tension until rupture, on a 2024-T3 aluminum alloy sheet. Initial crack sizes and orientations in the fatigue specimens were found to be randomly distributed. Acoustic emission was continuously monitored during the tests. Every few hundred cycles, the acoustic signal having the highest peak-amplitude, was recorded as an extremal event for the elapsed period. This high peak-amplitude is related to a fast crack propagation rate through a phenomenological relationship. The extremal peak amplitudes are shown by an ordered statistics treatment, to be extremally distributed. The statistical treatment enables the prediction of the number of cycles left until failure. Predictions performed a posteriori based on results gained early in each fatigue test are in good agreement with actual fatigue lives. The amplitude distribution analysis of the acoustic signals emitted during cyclic stress appears to be a promising nondestructive method of predicting fatigue life.     

Acoustic emission during fatigue of Ti-6Al-4V: Incipient fatigue crack detection limits and generalized data analysis methodology

The fundamentals associated with acoustic emission monitoring of fatigue crack initiation and propagation of Ti-6Al-4V were studied. Acoustic emission can detect and locate incipient fatigue crack extensions of approximately 10 m. The technique therefore can serve as a sensitive warning to material failure. There are three distinct stages during which acoustic emission is generated. These stages are: crack initiation, slow crack propagation and rapid crack propagation. The distinction between the stages is based primarily on the rate of acoustic emission event accumulation. These three stages of acoustic emission correspond to the three stages of the failure process that occurs during fatigue loading. That is, changes in acoustic emission event rate correspond to changes in crack extension rate. Acoustic emission event intensities are greater during crack initiation than during slow crack propagation and greatest during rapid crack propagation. In a given fatigue cycle, event intensities increase with increasing stress and most high-intensity events occur near or at the maximum stress. Acoustic emission may therefore be used with confidence to detect, monitor and anticipate failure, in real-time.     

Cluster analysis of acoustic emission signals in pitting corrosion of low carbon steel

The pitting corrosion characteristics of low carbon steel specimens are studied by acoustic emission (AE) and electrochemical techniques, in a 3.0 wt.% NaCl solution acidified to pH 2.0. The acoustic emission signals generated by pitting corrosion are classified based on multiple acoustic emission parameters using K‐means clustering algorithm, then each classified signals are analyzed by acoustic emission parameters correlation plot and distribution with time. Furthermore, each acoustic source characteristics is extracted using Gabor wavelet transform (WT) in the time and frequency domain. An error back propagation (BP) artificial neural network (ANN) is trained according to the classified signals, so as to successfully identify the acoustic emission signals from parallel experiments. Experimental results show that the hydrogen bubble activation, oxidized film rupture and pit growth are typical acoustic emission sources in pitting corrosion process, which can be effectively classified by cluster analysis and recognized by back propagation neural network. The data gathered from laboratory tests combined with the real data from acoustic emission on‐line storage tank floor inspection can help to evaluate the bottom corrosion severity and interpreter the corrosion source, further to make the on‐site testing more reliable and reduce the risk.     

基于声发射技术的热障涂层损伤行为

热障涂层以优异的隔热、耐磨和耐蚀性而被广泛应用于航空涡轮发动机中。由于热障涂层体系内部结构复杂,服役环境苛刻,导致其失效不可预测。热障涂层系统内的表面开裂和界面分层是限制热障涂层长时间使用的瓶颈问题,且热障涂层的过早剥落失效会导致合金基体暴露在高温燃气中,这可能引起灾难性的后果。针对涂层的裂纹扩展行为,最重要也最直接的研究方法就是对热障涂层的整个失效过程进行实时无损检测,为寿命预测提供最直接的证据。声发射技术是一种实时动态的无损检测方法,可直接检测热障涂层失效过程中的裂纹扩展行为,因此在热障涂层失效检测领域得到了广泛的应用。然而,造成热障涂层损伤失效的因素较多,如失效机理复杂、失效形式多样,以及声发射信号本身的随机性和不可逆性,使得利用声发射技术检测热障涂层失效整个过程的研究还不够全面。目前,已通过声发射技术的参数分析和波形分析实现了对热障涂层损伤失效的定性、定量和定位分析,并对涂层寿命进行了预测。参数分析是以多个简化的波形特征参数来表示声发射信号的特征,即对一些特征量进行统计的过程,如能量、频率、幅度等。采用声发射特征参数法可定量评估热障涂层的损伤程度并对涂层的寿命进行预测。目前人们从连续损伤累计、某一特定参量变化等多个角度预测热障涂层的寿命,但是各种寿命预测模型主要是根据实验结果的经验或半经验公式,随着热障涂层的发展以及对热障涂层失效机理认识的不断加深,寿命预测模型也在不断发展与完善。波形分析是通过对声发射信号的时域波形或频谱特征分析来获取缺陷信息的一种信号处理方法。从理论上讲,波形分析应当能给出任何所需的信息,因而波形也是表达声发射源特征最精确的方法,目前主要通过小波变换把声发射波形信号从时域变换到频域,进而识别其损伤模式并实现声发射源的定位。本文对声发射技术进行了简要的介绍,总结了声发射技术参数分析和波形分析在热障涂层损伤模式识别、损伤位置的定位、损伤程度的定量评估和剩余寿命预测方面的研究进展,指出了当前研究中存在的问题并对其下一步的发展进行了展望。     

基于时序分析的拉深件成型裂纹的特征参数提取

研究金属拉深制件声发射信号特征参数的提取对判断成型制件的质量尤为重要。研究采用时序分析和MATLAB对经过小波包分解的声发射信号进行特征参数的提取。在此过程中利用时间序列的自相关系数和偏相关系数的拖尾性及截尾性来判断模型的类型;采用FPE (final prediction error,最终预测误差0029准则对模型进行定阶;以最小二乘估计法对模型的参数进行估计。最后,根据所建立的自回归谱模型提取声发射信号的特征参数。研究结果是通过此方法成功提取到了金属拉深制件声发射信号的特征参数及其分布图,为成型制件质量的判断提供了有利的依据。     

Prediction of low-cycle fatigue-life by acoustic emission—2: 7075-T6 aluminum alloy

Low cycle high stress fatigue tests were conducted by tension-tension on an Alclad 7075-T6 aluminum sheet alloy, until rupture. Initial crack sizes and orientations in the fatigue specimens were randomly distributed. Acoustic emission was continuously monitored during the tests. Extremal peak-amplitudes, equivalent to extremal crack-propagation rates, are shown to be extremally Weibull distributed. The prediction of the number of cycles left until failure is made possible, using an ordered statistics treatment and an experimental equipment parameter obtained in previous experiments (Part 1). The predicted life-times are in good agreement with the actual fatigue lives. The amplitude distribution analysis of the acoustic signals emitted during cyclic stress has been proven to be a feasible nondestructive method of predicting fatigue life.     

Acoustic emission detection of fatigue cracks in wind turbine blades based on blind deconvolution separation

The occurrence and expansion of fatigue cracks in large wind turbine blades may lead to catastrophic blade failure. Each fatigue phase of a material has been associated with a typical set of acoustic emission (AE) signal frequency components, providing a logical base for establishing a clear connection between AE signals and the fatigue condition of a material. The relevance of efforts to relate recorded AE signals to a material's mechanical behaviour relies heavily on accurate AE signal processing. The main objective of the present study is to establish a direct correlation between the fatigue condition of a material and recorded AE signals. We introduce the blind deconvolution separation (BDS) approach because the result of AE monitoring is usually a convoluted mixture of signals from multiple sources. The method is implemented on data acquired from a fatigue test rig employing a wind turbine blade with an artificial transverse crack seeded in the surface at the base of the blade. Two different sets of fatigue loading were conducted. The convoluted signals are collected from the AE acquisition system, and the weak crack feature is extracted and analysed based on the BDS algorithm. The study reveals that the application of BDS‐based AE signal analysis is an appropriate approach for distinguishing and interpreting the different fatigue damage states of a wind turbine blade. The novel methodology proposed for fatigue crack identification will allow for improved predictive maintenance strategies for the glass‐epoxy blades of wind turbines. The experimental results clearly demonstrate that the AE signals generated by a fatigue crack on a wind turbine blade can be synchronously separated and identified. Characterizing and assessing fatigue conditions by AE monitoring based on BDS can prevent catastrophic failure and the development of secondary defects, as well as reduce unscheduled downtime and costs. The possibility of using AE monitoring to assess the fatigue condition of fibre composite blades is also considered.