【作者】 韩敬华；

【导师】 单平；

【作者基本信息】 天津大学 ， 材料加工工程， 2009， 博士

【摘要】 接触电阻是铝合金电阻点焊焊接质量不稳定、电极烧损严重寿命短等问题的重要影响因素。现有的接触电阻模型难以准确反映出焊接表面状况、焊接材料的物性参数和点焊参数对接触电阻的影响,因此本文以接触电阻为核心进行了一系列研究。主要研究内容如下。首先,采用改进了的Weierstrass-Mandelbrot分形函数对电极和工件表面形貌进行了表征,根据分形理论建立了点焊前期的弹塑性接触模型,分析了点焊前期的微观接触行为,考察了各种物性参数对接触行为的影响。分析结果显示:分形维数表征了不同加工和处理过程产生的工件和电极的表面状态;点焊中所使用的铝合金板和电极表面都具有多重分形特征;对于给定的载荷和界面,当分形维数从1.30增大到1.90时,接触率在分形维数D = 1.50时达到最小;对于铝合金来说,电极/工件间接触率小于工件/工件间接触率。其次,在接触分析的基础上,构建了无氧化膜时接触电阻网络。将康托分形模型应用于氧化膜的开裂,建立了氧化膜存在时接触载荷作用下铝合金接触电阻分形模型,推导出氧化膜存在情况下的接触电阻公式。采用单板法和双板法对建立的分形接触电阻模型进行了实验验证。验证结果显示:建立的模型反映了实际电阻随接触压力的负指数变化,与实验测量的电阻曲线吻合。深入分析发现,该模型是前人建立的二级和三级电阻模型的一般情况,具有一定的普遍意义。结合接触分析和建立的氧化膜条件下接触电阻模型分析了点焊的形核过程,结果发现点焊熔核的起始形核具有离散性。最后,采用有限元方法模拟了计算所得接触电阻条件下直流点焊熔核的温度场和尺寸变化,并利用人工神经网络研究了焊接参数、熔核尺寸与熔核拉剪性能之间的非线性关系。研究结果说明:点焊熔核在形核时首先在径向方向增长,然后在厚度方向增大;样条插值拟合方法不能揭示出焊接参数、熔核尺寸与接头拉剪强度之间的非线性关系;BP神经网络是描述焊接参数、熔核尺寸和拉剪强度三者之间关系的有力工具,采用两个三层BP神经网络对熔核的拉剪强度进行了预测,预测误差在8%左右,预测结果可信。更多还原

【Abstract】 Contact resistance is one of the main factors which influence the quality of weldment, the burning loss degree and life of electrode, etc. The models of contact resistance built as yet are difficult to describe the effect of surface morphology of workpiece and electrode, the physical properties of welding materials and the welding parameters on the contact resistance. To build a contact resistance model including the factors mentioned above, a series of studies around the contact resistance are performed. The main results obtained in this thesis are listed as follows.At first, the revised Weierstrass-Mandelbrot fractal function is utilized to characterize the surface morphology of electrode and workpiece, according to the fractal theory, the elastoplastic contact model at the early stage of resistance spot welding of aluminum alloy is setup, based on which the microcontact behaviors at the early stage of resistance spot welding of aluminum are analysized, but also the influences of all kinds of material physical parameters on the microcontact behavior are studied. It is found that the parameters of fractal dimension and fractal characteristic length characterized the surface morphologies of workpiece and electrode produced by different processing methods. The surfaces of electrode and workpieces are multifractal. For given load and interface, when fractal dimension changes from 1.30 to 1.90, the contact rate reaches the least value at 1.50. For aluminum alloy, the contact rate of Electrode/Workpiece is less than that of Workpiece / Workpiece.Secondly, based on the results of contact analysis, the network of contact resistance without oxide film on the surface is constructed, the Cantor set is applied to the crack of oxide film, the Cantor crack model of oxide film is presented, the fractal contact resistance model of aluminum under the contact load in which the effect of oxide film is considered is setup, the contact resistance formula with the oxide film is derived. The fractal contact resistance model is verified with the mono-plate and bi-plate methods, the results show that the resistance curve of this model agrees well with that of the experiment, and it reflects the negative exponent relationship between contact resistance and contact load. After detailed analysis, it is found that the model is the general case of two-level and three-level contact models built before, the model reflects a general law. The electrical contact resistance built combined with the contact analysis are utilized to describe the formation of nugget, it is found that the nugget initializes on many contact spots at the same.Finally, with the contact resistance caculated, the temperature field and the size of nugget are simulated using finite element method, the relationships between welding parameters, nugget size and the tension-shear strength are studied by artifical neural network. The results show that the nugget grows radially at first, and then expands in the direction of thickness. The spline fitting method can’t reflect the nonlinear relationship between the welding parameters, size of nugget and the tension-shear strength of nugget, but the BP artifical neural network can. The tension-shear strength is predicted using two BP neural networks with three-layers, the predictive error is about 8%, it means that the predictive result is reasonable, and the BP network can be used to predict the tension-shear strength and the size of the spot nugget.更多还原