电阻点焊接头熔核面积的预测

A prediction of nugget area in RSW

以电阻点焊接头表面的数字图像作为信息源,探索了一种新的点焊接头质量无损检测方法。首先通过图像特征分析,将焊点表面图像划分为三个特征区域,提取三个同心圆的面积作为表征接头熔核面积的特征参量。其次,以提取的特征值作为输入向量,以接头横截面处的熔核面积作为评判标准,建立点焊接头熔核面积的SVM(支持向量机)等级分类模型。实验结果表明,该模型可以有效地对熔核面积进行预测分类,实现了对电阻点焊接头质量的无损评判,经验证其准确率可达96.667%。

A new non-destructive method was explored to monitor the joint quality based on digital images of resistance spot welding joints＇ surface.First ,welding joints＇ surface images were divided into three characteristic zones from which areas of three centric circle we~ extracted as characteristic parameters to characterize joints＇ nugget area.Then SVM （support vector machine） model of classification was established while the characteristic parameters were input vectors and the nugget area of joints＇ cross-sectional was considered as the evaluation criteria.Experimental results show that the model can predict and classify the nugget area effectively.It achieved the non-destructive evaluation of solder joints＇ nuclear area eventually.The accuracy rate of the model came up to 96.667%.