Droplet Transfer Pattern Recognition by SVM Based on Arc Acoustic ARMA Bispectrum Analysis

In order to do rapid recognition for droplet transfer in gas metal arc welding simply and effectively,welding arc sound was used to do droplet transfer pattern recognition.Through obtaining welding arc sound signals of different droplet tansifer,this paper analyzed welding arc sound of different droplet transfer by ARMA bispectrum and extracting eigenvector,then did pattern recognition to the obtained eigenvector by support vector machine(SVM).It can succeed in recognizing kinds of types of droplet transfer.