【作者】 朱荣华；

【导师】 刚铁；

【作者基本信息】 哈尔滨工业大学 ， 材料加工工程， 2013， 博士

【摘要】 随着高速铁路的不断发展，铝合金在列车结构如牵引梁、枕梁等构件中得到广泛引用。在列车高速运行过程中，铝合金结构由于受到交变载荷的作用，易产生疲劳损伤，从而危及列车的安全运行。目前，常规无损检测方法需要在列车停运状态下对铝合金结构进行离线的检测，因而无法对高速运行过程中的铝合金结构进行在线状态监测。与常规的无损检测技术相比，声发射技术具有对活动性缺陷敏感、可对大型结构快速整体检测等优点，但是目前声发射技术在列车中的检测应用还处于起步阶段。由于对列车用铝合金损伤的声发射特征以及列车运行过程中的噪声信号还缺少认识，因此研究其损伤的声发射特征，以及在运行噪声环境下提取铝合金损伤的声发射信号成为目前列车声发射监测急需解决的问题。针对列车声发射监测存在的问题，本文开发了一套显微图像原位观测系统，该系统可以对材料损伤过程的行为进行连续图像监测及采集，可为声发射信号的分析提供技术手段。采用声发射系统与开发的显微图像原位观测系统同步监测列车用铝合金母材及焊缝静载损伤过程，分析了声发射信号时域、频域、时频域的特征，建立了声发射特征与静载损伤裂纹扩展之间的关系，并通过断口分析确定了静载损伤过程中的声发射源，研究结果显示：声发射能量、峰值频率及质心频率的变化可有效监测铝合金微裂纹的萌生。铝合金结构在列车服役过程中受交变载荷的作用易产生疲劳裂纹，而疲劳裂纹扩展速率是评估铝合金结构剩余寿命的基本数据。在实验室条件下，采用声发射及自制的显微图像原位观测系统监测了铝合金疲劳过程，研究了铝合金母材及焊缝疲劳损伤过程中裂纹萌生、扩展及相应的声发射特征，并通过疲劳断口分析了声发射的产生源。研究结果表明：铝合金疲劳损伤过程中的声发射存在三个不同的变化阶段，分别与疲劳裂纹扩展的三个阶段相对应。在疲劳裂纹萌生阶段，声发射特征值的变化对应的疲劳循环次数早于显微图像原位观测系统首次发现疲劳微裂纹的循环次数，因此声发射特征值可作为疲劳裂纹萌生的指标。建立了声发射计数率与疲劳裂纹扩展速率之间的关系，通过声发射监测得到的特征值以及建立的关系可以计算出疲劳裂纹的扩展速率，从而评估铝合金结构的健康状态，避免了列车结构中应力强度因子的测量困难。为了分析列车运行过程中噪声的特征，开发了一套小型现场加载系统，在列车运行过程中，采用该系统对列车用铝合金进行加载，来模拟列车服役过程中的声发射信号，同时利用数据采集系统实时获取噪声环境下的声发射信号，为后续的信号分析与处理提供了基础数据。通过对列车静止和运行过程中，铝合金不加载与加载过程产生的声发射信号对比分析，得到了列车运行过程中的噪声信号类型。对采集的噪声与裂纹声发射的混合信号，首先采用时差法分离出噪声与裂纹声发射两个信号集合，分别在时域、频域、时频域分析了噪声信号的特征，并提取了噪声与裂纹声发射信号在上述值域的特征值。对提取的特征值采用欧氏距离评价准则进行了优化，得到最佳特征值子集。在此基础上构建了BP神经网络分类器，对噪声与裂纹声发射信号样本进行了智能识别，有效分离了噪声信号，从而降低了噪声信号对声发射检测结果的影响。更多还原

【Abstract】 With the development of high-speed railways, aluminum is widely used in trainstructures, such as the traction beams, corbels and other components. When the trainruns at high speed, fatigue damage will be caused by alternating load in thealuminum structure, which endanger the safe operation of the train. Currently,non-destructive methods need to stop the train and test the body of the train off-line,so they are unable to monitor the body parts during operating condition. Comparedwith conventional non-destructive testing methods, acoustic emission technology issensitive to active defects, and can detect the overall structures fast and efficiently,but the application in trains is still in its infancy. Due to the lack of awareness of AEcharacteristics of aluminum alloy and noise signals in running trains, it is urgent toresearch acoustic emission signals from damage of aluminum and extract them fromnoisy environment when acoustic emission technique used in trains.For the problems of acoustic emission monitoring used in trains, a new in-situmicroscopic observation method was proposed in this paper, which can monitor thematerial damage process and capture damage images continuously, and it willprovide technical assistance to the analysis of acoustic emission signals.Acoustic emission and the in-situ microscopic observation system were used tomonitor aluminum base metal and weld during static damage process. The time,frequency and time-frequency domain characteristics of acoustic emission signalswere analyzed to establish the relationship between the characteristics and staticdamage behavior. The acoustic emission sources were identified by the fractureanalysis. The results show that acoustic emission energy, peak and centroidfrequency were effective indicators to monitor the crack initiation of aluminum.Fatigue cracks are easy to cause by alternating loads in aluminum structureduring the service process of trains, and fatigue crack growth rate is the base data toassess damage degree of aluminum structures. Acoustic emission features of crackinitiation and propagation in fatigue damage process of the aluminum base metaland weld were studied under laboratory conditions, and fatigue processes weremonitored by acoustic emission and in-situ microscopic observation system. Thesources of acoustic emission were analyzed by fatigue fracture. The results showthat fatigue damage of aluminum has three different acoustic emission stages, whichcorresponds to the three stages of fatigue crack growth respectively. In the fatiguecrack initiation stage, number of fatigue cycles corresponding to the change ofacoustic emission features was earlier than the number when micro-crack was firstdiscovered by the in-situ microscopic observation system. Therefore the acoustic emission features can be used as indicators of fatigue crack initiation. Relationshipbetween the acoustic emission count rate and fatigue crack growth rate wasestablished. The fatigue crack growth rate can be calculated through the monitoredacoustic emission features and the established relationships. Thus the health statusof aluminum structures was assessed, and the problem of difficult measurement forthe stress intensity factor in actual structure was avoided.In order to analyse the noise features in the moving train, a small in-siteloading system was developed to load aluminum alloy in a running train, whichsimulated acoustic emission signals in the service process of the train. Dataacquisition system was used to get real-time acoustic emission signals under noisyenvironments, which were the basic data provided for subsequent signal analysisand processing. Types of noise signals were analyzed through comparative analysisof acoustic emission signals obtained on the static and operational train, whenaluminum is not loaded and loaded. For the collected mix signal that contain noiseand acoustic emission signals from crack propagation, the time difference methodwas used to separate two signal sets of noise and acoustic emission signals fromcrack propagation. The characteristics of noise signals in time and frequency domain,time-frequency domain were analyzed, then the features of the noise and AE signalsfrom crack propagation in the above range were extracted. The extracted featurevalues were optimized to get the best feature subset using the Euclidean distanceevaluation criteria. On this basis, BP neural network classifier was constructed toidentify samples of noise and acoustic emission signals from crack propagation,which effectively separated the noise signals, thus lay a foundation for the acousticemission technique to monitor the aluminum structures used in trains.更多还原