【作者】 苏杭；

【导师】 常云龙；

【作者基本信息】 沈阳工业大学 ， 材料加工工程， 2007， 硕士

【摘要】 电阻点焊是一种重要的焊接方法，广泛应用于汽车、航空航天等工业。随着电阻点焊应用领域的不断扩展和深入，对电阻焊控制系统提出了更高的要求。但由于电阻点焊是一个高度非线性、多变量和强耦合的复杂过程，点焊质量受到多种因素的影响，传统的采用单一参数反馈质量控制方法存在很大局限性。基于此本文设计了一种以89C51单片机为核心的双处理器微机控制系统，合理分配并细化微处理器的任务，提高系统的实时控制能力，为多参数综合质量监控以及智能控制的实现奠定基础。设计了双处理器的硬件电路及控制系统，对时钟电路模块、A／D、D／A转换电路模块、键盘及显示电路模块等进行了详细分析，并编写了焊接主程序、双机通讯程序、键盘及显示程序、AD／DA转换程序、焊接电流PID控制算法等程序。设计了电极位移测控系统，利用体积小、分辨率高的光栅传感器对点焊过程的电极位移进行实时检测；利用峰值角法计算电流有效值。以RBF神经网络为基础，建立了以焊接电流、焊接时间、电极压力、板厚作为神经网络的输入，熔核直径作为输出的神经网络模型，实现了对点焊工艺参数的优化设定。采用正交试验方法，针对生产中常用的板材低碳钢在不同的焊接参数下进行等厚双板焊接，通过多组试验获得了点焊工艺参数的最优组合，并对所建立的神经网络模型进行了验证。点焊试验结果表明，神经网络模型对熔核尺寸的预测误差小于8％，说明建立的神经网络模型是正确的，预测结果准确可靠。对熔核尺寸的预测精度可以满足实际生产的要求，对于提高点焊质量具有一定的理论与实际应用价值。更多还原

【Abstract】 Resistance spot welding is a kind of important welding method which is widely used in automobile, aerospace and so on. With the continuous expansion of its application fields, the higher requirements of control system are put forward. However, the resistance spot welding is a complex process which is highly-nonlinear, multi-variable and strong coupling. The quality is affected by various factors and the traditional quality control methods using a single-parameter feedback have limitations and drawbacks to some extent. For this reason, the double processors control system is proposed in this paper. The 89C51 single chip is used as kernel of the system, and the real-time performance is increased. The multi-parameters fusion intelligent control can be realized through rational assignment of the duty of double processors.The hardware circuit and control system are designed. The clock circuit module, A/Dand D/A converter circuit module, keyboard and display circuit are analyzed in detail. The welding sequence, double processors serial communication, keyboard and display, A/D and D/A converter process, the digital PID algorithm are programmed. The electrode displacement system is designed, the grating sensor which has small size and high resolution is adopted to measure the displacement during spot welding. The current effective value is calculated by the method of peak value angel. The mathematic model is established based on RBF neural network, using the parameters of welding current, welding time, electrode pressure and board thickness as input of neural network system, and the nugget diameter as output of it, from which the optimized parameters of spot welding is realized. The orthogonal experiment is carried out on low carbon steel plate in different technological parameters for many tests, by which the optimal combination of technology parameters is abtained. The neural network model is also verified through experiments.The spot welding results show that the estimated error is less than eight percent, which expresses that the established model is precise, so the estimated result is reliable. The predicted precision can meet the actual production request. The research results have great values in theory and application for spot welding.更多还原