基于信号特征提取的电阻点焊质量在线评判

On-line quality estimation of resistance spot welding based on extraction of signals feature

关键词:  电阻点焊  特征提取  数据泛化  Hopfield神经网络  模式识别

Key words:  resistance spot welding  feature extraction  data generalization  Hopfield neural network  pattern recognition

摘要:

以焊点接头强度作为焊点质量评判的指标,通过对点焊过程焊接电流、动态电阻、电极位移信号的同步采集和特征分析,提取若干特征参量监测点焊过程,依据特征参量与焊点接头抗剪强度间的相关分析结果,选取来自不同监测信号的7个特征参量建立了表征点焊过程的特征模式,并将此转化为计算机可以识别的模式矩阵,同时以焊接电流参数为模式分类的依据,建立不同模式矩阵类别和焊点接头抗剪强度之间的映射,将模式矩阵作为Hopfield神经网络的记忆样本存储于网络,利用网络联想记忆的功能实现对未知样本点焊过程的模式识别,进而实现点焊质量的评判。网络测试结果表明,利用Hopfield网络进行焊点质量在线评判可以得到满意的效果。

Abstract:

For estimating joint quality of resistance spot welding,joint tensile-shear strength was taken as a kind of evaluation criterion. Welding current,dynamic resistance and electrode displacement signals were simultaneously monitored and collected in welding process.Through character analysis,several characteristic parameters relating to the weld quality were extracted from the three signals.At the same time,based on the correlation analysis results between the parameters and weld strength,7 characteristic parameters were selected to taking as the characteristic pattern of welding process.All characteristic patterns were converted into two-dimension pattern vectors,which were accessed by the computers.Then those patterns were classified according to different welding current.At last,a kind of estimating model of Hopfield neural network was established on the mapping of different pattern vectors and joint strength.Network test results indicated that Hopfield neural network model could get satisfactory effect in resistance spot welding(RSW) quality online evaluation.