熔滴过渡光谱信号模式识别分类器的设计

以熔化极气体保护焊电弧光谱信号作为样本,设计了一种熔滴过渡模式识别分类器.首先对光谱信号进行预处理并抽取了多个关键性的特征参数,通过降维分析得到一组新的特征向量.随后建立了相应的识别函数和最小距离法分类器.最后利用检验样本对分类器的性能进行了检验和评价.判别结果表明,利用该分类器能够较好地对MIG焊和C02焊熔滴过渡类型进行自动识别,具有较高的准确性和识别精度,为实现熔化极气体保护焊熔滴过渡自动控制奠定了基础.

Design of the pattern recognition classifier of the spectrum signal of droplet transfer

A pattern recognition classifier of droplet transfer mode was designed using the arc spectrum signal of gas metal welding as samples. The spectrum signal was pretreated and several key characteristic parameters were extracted and then a set of new feature vector was obtained by reducing the dimensions.Corresponding recognition function and a minimum distance classifier were constructed and finally the test samples were used to test and evaluate the property of the classifier.Results showed that droplet transfer modes of MIG and CO_2 welding were recog- nized automatically with high veracity and identiable accuracy,which provided the basis for automatically controlling the metal gas welding droplet transfer.