论文部分内容阅读
针对电阻点焊压痕深度的检测存在离线、滞后等问题,提出了一种基于电极位移信号特征提取的人工智能在线预测压痕深度的实现方法。首先,采用搭建的计算机激光测量系统探索了焊点压痕深度测量方法,确定了以多次测量的平均值hT作为压痕深度的实际评定值。其次,通过对熔核形成过程、电极位移信号与焊点压痕深度的相关性研究,确定了焊接电流I、电极压力F、以及从电极位移信号中提取的特征参量h作为压痕深度的表征参量。最后,采用压痕深度的表征参量作为输入向量,以测定的焊点实际压痕深度hT作为目标向量,建立了SVM(supportvectormachine)回归预测模型。实际测试表明,模型输出的压痕深度预测值和实际测定值间的线性相关度达到了91.18%,通过实时监测熔核形成过程,可以实现焊点压痕深度的预测。
In order to solve the problem that the indentation depth of resistance spot welding is off-line and hysteresis, a method of on-line artificial indentation depth prediction based on the feature extraction of electrode displacement signal is proposed. First of all, using the built-in computer laser measurement system to explore the solder joint indentation depth measurement method to determine the average of multiple measurements hT as indentation depth of the actual evaluation value. Secondly, through the study of the correlation between nugget formation process, electrode displacement signal and indentation depth, the welding current I, electrode pressure F, and the characteristic parameter h extracted from the electrode displacement signal were determined as indentation depth Parameter. Finally, using the indentation depth characterization parameters as input vector, and using the measured actual indentation depth hT as the target vector, a SVM (support vector machine) regression prediction model is established. The actual tests show that the linear correlation between the predicted indentation depth and the actual measured value reaches 91.18%. The real-time monitoring of nugget formation can predict the indentation depth.