人工神经网络技术在点焊质量控制中的应用研究

ARTIFICIAL NEURAL NETWORKS APPLIED TO THE QUALITY CONTROL IN ALTERNATING CURRENT RESISTANCE SPOT WELDING

利用人工神经网络技术对交流电阻点焊的多个动态电参数进行融合处理,建立起以交流点焊过程中动态电参数作为输入空间;以熔核尺寸为输出空间,可用于实时在线检测和预测的低碳钢点焊质量监测系统。所建监测系统的熔核直径的平均预测误差小于5%,熔核高度的平均预测误差小于8%,完全可以满足工程实际的需要。

Several of the dynamic electrical parameters of alternating\|current resistance spot welding are blended by use of artificial neural networks,and a monitor system of spot welding quality of mild steel is established in this research.In this system,the dynamic electrical parameters are used as input space and the sizes of nugget are used as output space.The system can be used for detecting the quality and forecasting the size of nugget on real time during resistance spot welding. The average forecasting error of diameter of nugget is less than 5% and the average forecasting error of height of nugget is less than 8% in this monitor system. The system can satisfy the actual need of engineering completely.