【作者】 罗震；

【导师】 单平；

【作者基本信息】 天津大学 ， 材料加工工程， 2003， 博士

【摘要】 点焊是现代汽车制造工业中重要的工艺方法，它具有生产效率高、易于实现自动化而被广泛地应用于汽车的焊装生产线上。近年来，汽车轻量化的研究把铝合金作为重要材料，由于铝合金的焊接性较差、点焊过程中影响熔核的因素多，使铝合金点焊质量受到影响，如何保证铝合金点焊质量一直是国内外焊接界学者们致力于研究的重要课题之一，本文对铝合金点焊质量监控作了以下三个方面的工作：第一，对铝合金点焊的初始接触电阻进行了深入研究通过初始接触电阻的线性方程组建模与数理统计分析，得到接触电阻的数值解，并且提出了初始接触电阻的概率分布函数。众所周知，不同电极压力下，初始接触电阻值变化很大，尤其是在电极压力较低的情况时，如0.8mm厚铝合金5052其概率密度函数呈现尖峰和单边倾斜的形状。笔者首次采用稳定分布来描述接触电阻，将不同压力下的初始接触电阻分布统一起来，因此可以通过稳定分布的理论成果来对初始接触电阻进行深入的研究，为初始接触电阻及点焊质量控制研究提供了理论基础。第二，对铝合金点焊的质量监控参数进行研究焊接电压信号的小波包能谱研究。对铝合金点焊焊接电压信号进行小波包变换并计算其能谱，研究了能谱与熔核质量之间关系。分析与试验结果表明，焊接电压的小波包能谱是一种比较理想的质量监控参数。点焊过程声信号的研究。点焊过程声信号研究是建立在熔核向外界释放的声能量与熔核的质量之间的关系上，笔者将声音信号进行了时频域分析，得到了铝合金点焊过程中以前未能提取出来的细节信息，增加了对熔核变化过程的了解；为了能给控制器提供控制参量，笔者应用RENYI熵作为点焊过程的特征量，并对RENYI熵与熔核面积之间的关系进行了讨论，计算结果表明：RENYI熵能很好地描述熔核形成过程。点焊电极位移信号研究。位移信号能直接反映熔核形成过程的动态变化，笔者通过试验证明了电极位移参数与铝合金点焊熔核之间有着较强的相关性。为此，笔者建立了以时间序列为基础的AR（n）自回归模型，计算了电极位移、阻尼比、振动频率和功率谱，并且对它们与熔核质量之间关系作了研究，同时将电极位移、电极速度和位移振动信号第一峰值能量作为铝合金点焊质量的监控特征量。第三，对铝合金点焊质量的控制方法进行研究铝合金点焊过程是多因素影响的复杂过程，要想得到高质量的监控效果，需要采用多物理量的信息融合技术，同时控制方法也要进行融合，以此来适应<WP=7>精确的点焊质量要求。本文对各种检测与控制方法进行了分析，提出了基于遗传算法的数据融合神经网络技术作为点焊质量监控的方法。同时，笔者创建了“熔核面积”与“点焊电流”之间的数学模型，为铝合金点焊的质量控制奠定了基础。更多还原

【Abstract】 Resistance spot welding (RSW) is a major sheet-metal joining means for auto body assembly in automotive industry. The advantage of RSW is its inherently properties of the high productivity and the efficient robotization. Aluminum alloys is one of the valuable lightweight materials in the research of the weight reduction. With natures of multivariate, complication and less weldability, experts have been focusing their efforts on monitoring and control in order to improve spot weld quality over the years. In my dissertation, three parts of the work are studied in aluminum alloys spot welding of monitoring and control. First. Physical model of currentless contact resistance (CCR) has been established and CCR is calculated. The probability density function(PDF) of CCR has been put forward to investigate implicit contact resistance properties. It is well-know that CCR is determined by pressure of the electrode, especially in low pressure case, PDF of CCR has skewness and kurtosis. So the stable distribution has been employed firstly to capture the PDF of CCR in different pressure of electrode and the test shows that stable distribution can constitute a class of CCR distribution. The stable distribution developed in this study is primarily targeted to offer a new method studying contact resistance.Second. Research on monitoring and control parameters of weld quality in aluminum alloys spot welding.Research on wavelet packet decomposition and its energy spectrum using voltage signals. Wavelet packet decomposition and its energy spectrum deals with the signal of voltage and its relationship with weld quality. The result demonstrates that wavelet packet energy spectrum is a desirable approach of monitoring and control parameters in aluminum alloys spot welding. Research on the sound signals. The research is based on the fact that spot welding quality has correlation with sound energy from nugget growth in the time-frequency domain. In particular, the difference in sound emission of big nugget size and small nugget size is distinguishable from the time-frequency distribution.Time -frequency analysis has been used to diagnose the sound signals. As a<WP=9>result, some details, which are not discovered before, now show themselves in probability distribution. These details enhance the understanding of physical process of the nugget growth during spot welding.In order to extracted information from signals collected in the process, Renyi entropy is selected as feature extraction variables from the process of spot welding. Discussion of the relationship between Renyi entropy and nugget area has been presented. At the same time, the experiment shows that Renyi entropy can characterize the nugget growth.Research on electrode displace signals. The signals of electrode displace can show the nugget growth. The designed experiment demonstrates that there is considerable relationship between the mode parameters of electrode vibration and nugget growth. AR(n) autoregressive model has been applied to compute damp ratio, vibration frequency and power spectrum of electrode displace. And research on the relationship between these parameters and nugget quality are studied systematically. Thus the condition has been reached that the first peak in the energy spectrum of vibration signal can be used as the feature parameter for monitoring and control nugget growth. Third, Research on monitoring and control method.RSW is a complex process in which many electrical, mechanical and thermal process variables are involved, so weld quality depends on many variables. Due to the complex nature of RSW, information provided by one or two variables might be insufficient. Monitoring and control of high weld quality must be required, including not only multiple sensors data fusion but also combination of several control methods. A new monitoring and control means, based on genetic algorithm, data fusion and neural network, is presented through comparison with other methods in our research. In addition, feature-based process mo更多还原