【作者】 刘静；

【导师】 徐国成；

【作者基本信息】 吉林大学 ， 材料加工工程， 2015， 博士

【摘要】 伴随城市轨道交通的快速发展，作为主流车型之一的不锈钢城轨客车，其制造业亦得到了迅猛发展。为实现轻量化等设计目标，不锈钢城轨客车车体制造主要采用电阻点焊焊接技术。电阻点焊过程加热强度大、时间短，易产生飞溅、虚焊、脱焊等焊接缺陷，影响焊接质量的稳定性，进而影响不锈钢车体的制造质量。同时，点焊接头熔核的形成过程处于封闭状态不能直接观测，给点焊质量控制造成了很大困难。因此，点焊质量控制及评估技术的研究一直是被重点关注的领域，随着轨道车辆高速化及对制造质量要求的不断提高愈显其重要的意义。国内外常用的电阻点焊接头质量评估方法主要有两种：第一种方法是基于焊接过程参数的在线评估，主要是通过建立电阻点焊焊接电流、电极间电压、电极压力、动态电阻、电极位移等特征信息与接头质量关系的数学模型来预测焊点质量。然而，由于电阻点焊是一个非线性、多变量耦合作用且大量不确定因素相互影响的复杂过程，上述方法只能提供部分可靠的质量信息，并不能完全准确地对点焊质量进行评估。并且由于设定的试验条件苛刻或所建立的数学模型过于理想化，该方法的研究仍处于实验室阶段，无法应用于实际焊接生产；第二种方法是焊后破坏性检测，该方法需要对焊点进行剥离、扭转等破坏，获得熔核直径和接头强度的大致信息，再参照相关标准对点焊试件质量进行评估。目前企业在点焊工艺规范调试、点焊生产过程的自检和抽检中，多采取该方法。但该方法效率低，同时也造成了材料的浪费，增加了人力及生产成本，不适合作为电阻点焊接头质量的在线检测手段。近年来，许多学者开始研究电阻点焊质量的无损检测技术。相比于其他无损检测方法，超声波检测具有灵敏度高、检测范围广、检测速度快、安全性能好、易于实现现场操作等优点，备受各国学者关注。目前，电阻点焊质量超声波检测技术仍存在亟需解决的问题：如何区分熔核和塑性环，获得准确的熔核直径尺寸，提高检测精度；如何快速地对点焊质量进行评估，提高检测效率；如何实现点焊缺陷的智能化检测等，这些都大大地限制了该技术在实际工业生产中的应用。为了满足人们对车体安全、高质量的需求，推动电阻点焊超声波无损检测技术的发展，本文针对不锈钢电阻点焊质量超声波定量及智能化无损检测技术进行了研究，具有重要的理论和工程应用意义。本文成功研发了便携式专用电阻点焊全自动超声波检测系统，该系统不但操作简单、便携可靠，而且实现了对电阻点焊接头的接触式自动超声扫描检测及检测信号自动分析、C扫描图像自动生成等功能，为点焊质量超声波定量及智能化检测提供了可靠的数据来源。本文通过研究超声波在介质中的传播特性，构建了超声波在介质中传播反射系数模型，揭示了在一定的间隙厚度范围内，声压反射系数表现出依赖频率的变化特性，为检测点焊缺陷提供了思路，特别是为区分焊合与弱结合奠定了理论依据。分别在时域、频域和时频域上对点焊接头超声A回波信号进行了深入地分析，研究发现：超声回波信号特征值与点焊接头内部熔合状态存在一定的对应关系。通过提取能够表征熔核区域的特征信号，利用小波包变换分离出超声回波高频信号，并根据其频谱特性，可以对点焊熔核和塑性环进行区分，获得更加精确的熔核直径尺寸。本文选择了1000μm的扫描步长对点焊试件进行步进式机械扫描检测，利用小波包变换提取可以表征接头内部熔合状态的频率为16MHz的高频信号，获得整个扫描区域的低分辨率C扫描图像。采用双三次图像插值的方法对该C扫描图像进行细分重建，在增强图像分辨率的同时，检测效率提高了14倍，使实现点焊超声波C扫描快速检测成为可能，在实际工程应用领域具有非常积极的意义。针对插值后的C扫描图像进行图像增强、边缘检测处理，可获得清晰的熔核边界形貌，进一步突出了熔核特征，而且大大降低了运算数据量。通过等效直径算法程序，自动获取了点焊熔核尺寸，实现了点焊接头质量超声波检测的自动化、定量化的快速评估。通过金相试验验证，经检测获得的熔核直径尺寸和实测值吻合良好，且相对误差正态分布的均值和方差均很小，该检测方法具有非常高的精度和稳定性，可以用来评价电阻点焊质量，而且比单一的拉伸破坏性试验更加可靠。本研究利用有限元技术对不锈钢电阻点焊超声检测过程进行了仿真计算，揭示了超声波在点焊接头内部的传播特性及散射规律，为探索熔核边缘识别和熔核直径定量计算方法提供理论依据。针对点焊熔核中容易出现的气孔缺陷进行了模拟，研究不同深度的气孔缺陷对超声波传播特性的影响，并通过实验验证了所采用仿真方法及模型的正确性及有效性。为检测过程中波形分析提供了非常有用的信息，同时也为缺陷识别提供了检测样本。本研究利用所开发的超声检测系统对开焊、虚焊、缩孔及合格接头等不锈钢电阻点焊试件进行了检测，提取了能够反映不同点焊类别的特征信号，并通过BP神经网络对特征信号进行了计算机自动识别及分类，正确识别率达到了96%以上，实现了不锈钢电阻点焊质量超声波智能化检测。更多还原

【Abstract】 With the rapid development of urban rail transit, the manufacturing of stainless steelurban rail car which is one of the mainstream vehicle models, develops rapidly as well. Thejoining of the stainless steel car body mainly adopts the resistance spot welding technologyto achieve design goals of lightweight, etc. In the welding process, defects like spatter,incomplete weld or loose weld are easy to appear because of its high heating intensity andshort welding time, which affects the stability of welding quality and the quality of thestainless steel body. At the same time, the formation process of the spot welding nugget isin the closed state, which cannot be observed directly and cause great difficulty in the spotwelding quality control. Therefore, the technology of the quality control and evaluation inthe spot welding has always been paid close attention to. The significance of this studybecomes more important as the requirement of high-speed railway vehicles andmanufacturing quality are increasing continually.There are mainly two kinds of quality evaluation methods used commonly in theresistance spot welding at home and abroad. The first method is online evaluation based onwelding process parameters. It predicts the welding quality mainly through themathematical model which is established by the relationship between the joint quality andthe characteristic parameters like the welding current, the dynamic resistance, the voltagebetween electrodes, the electrode pressure or the electrode displacement. However, sincethe resistance spot welding is a nonlinear and multivariable coupling complex process witha large number of uncertain factors influencing each other, the method described above canonly provide some reliable quality information, but cannot be used to accurately evaluatethe welding quality. Because of the extreme operating conditions and almost ideal demandof the mathematical model, the research of this method is still in the laboratory stage, andcannot be applied to the actual welding production.The second method is destructive test after welding. This method need to peel and reverse the spot welds to obtain general information of the nugget diameter and jointstrength, and then to evaluate the welding quality with reference to relevant standards. Atpresent, this method is mostly adopted in enterprises to debug the spot weldingspecifications and carry out the random sampling inspection in the production process. Butthis method is of low efficiency, and it also wastes materials and increases the human andproduction costs. So, this method is not suitable as the on-line detection means to evaluatethe resistance spot welding quality. In recent years, many scholars are devoted to thenondestructive detection technology of the resistance spot welding quality. Comparing withother nondestructive detection methods, the ultrasonic detection draws more attention withthe advantages of high sensitivity, wide test range, fast test speed, good safety performance,and easy for site operation.At present, there are still many problems to be solved urgently in the ultrasonicinspection technology for the resistance spot welding, such as how to distinguish the coronabond from the nugget to obtain the accurate nugget diameter and improve the detectionaccuracy, how to quickly evaluate the spot welding quality and improve the detectionefficiency, and how to realize the intelligent detection of welding defects. These problemsgreatly limit the application of the technology in the actual industrial production. In order tomeet the demand of car body’s security and stability and promote the development ofultrasonic nondestructive detection technology for the resistance spot welding, this paperstudies the quantitative and intelligent ultrasonic detection for the resistance spot weldingof stainless steel, which has important significance in the theory and engineeringapplication.This paper makes a research on a portable spot welding automatic ultrasonic detectionsystem. The system is portable and reliable, and it is not only simple to operation but alsorealizes the function of self-adaption scanning detection, real-time signal acquisition andC-scan images automation generation without the probe submerged in the water, whichprovides the reliable data sources for the quantitative and intelligent ultrasonic detection forspot welding quality. In this paper, the ultrasonic reflection coefficient model in a medium is constructedthrough studying the characteristics of the ultrasonic propagation. It reveals that in a certaingap thickness range, the acoustic pressure reflection coefficient shows frequency-dependent,which provides a research thought for testing spot welding defects and especiallyestablishes a theoretical base for distinguishing the weak bonding. The study analyze theultrasonic A-echo signal of the spot welding joint in time domain, frequency domain andtime-frequency domain respectively and it is found that the ultrasonic echo signal has acertain corresponding relationship with the internal fusion state of the spot welding joint.The characteristic signal which can represent the spot weld is extracted and thehigh-frequency ultrasonic echo signal is separated using the wavelet packet transform toeffectively identify the nugget and the corona bond according to the spectrumcharacteristics, and the accurate nugget diameter is finally obtained.The stepping mechanical scanning detection is carried out on the spot weldingspecimens with the scanning step length1000μm. The16MHz high frequency signal whichcan represent the internal fusion state of the welding joint is extracted by the wavelet packettransform to get the low resolution C-scan images in the whole scanning area. The methodof the bicubic image interpolation, which subdivides and reconstructs the C-scan images,enhances the image resolution, and the detection efficiency is also improved14times. Itoffers the possibility of realizing the rapid spot welding ultrasonic C-scan detection and hasvery positive significance in the practical engineering applications. After the bicubic imageinterpolation of the C-scan images, the computer image enhancement and edge detectionprocessing are conducted to get the clear nugget boundary topography, which furtherhighlights the nugget features and greatly reduces the amount of computation. Then thespot welding nugget diameter is automatically obtained using the equivalent diameteralgorithm procedure, and the automatic and quantitative ultrasonic detection andassessment for the spot welding quality is finally realized. Through the metallographicexperimental verification, the testing values of the nugget diameter fit the actual measuredvalues well, and both the mean value and the variance of the normal distribution of relative error are very small. The detection method is of high accuracy and excellent stability. It canbe used to evaluate the quality of resistance spot welding and is more reliable than thesingle tensile destructive test.The simulated calculation is carried out on the ultrasonic detection process forstainless steel resistance spot welding by the finite element technology. The calculationreveals the ultrasonic propagation characteristics and the scattering regularity in the inner ofthe spot welding joint, which provides the theoretical basis for exploring the method ofnugget edge recognition and quantitative nugget diameter calculation. Then the weldingporosity which easily appears in the spot welding nugget is simulated, and the effect of itsdifferent depth on the ultrasonic propagation characteristics is studied. The correctness andvalidity of the simulation and the model are verified by the experimental method. Thesimulation method provides very useful information for the waveform analysis during thetesting process and offers experimental samples for the defect recognition.In this research, four kinds of resistance spot welding specimens, including loose weld,incomplete weld, shrinkage cavity and normal spot weld are tested by the developedultrasonic detection system. The characteristic signals which can reflect different spotwelding types are extracted. Characteristic signals of welding defects are automaticallyrecognized and classified on the computer by BP neural network, and the correctrecognition rate reaches more than96%, which realizes the intelligent ultrasonic detectionfor resistance spot welding quality of stainless steel.更多还原