A quality diagnosis method of GMAW based on improved empirical mode decomposition and extreme learning machine

Due to the non-stationary and nonlinear characteristics of arc signal in gas metal arc welding (GMAW), results in the difference of frequency distribution. In this study, a method for evaluate weld quality based on complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) and extreme learning machine (ELM) is proposed. First, the current signal is decomposed into intrinsic mode functions (IMFs) of different frequency bands by CEEMDAN, and then the energy entropy of IMFs is extracted. Because of the energy of each IMF under different weld quality is varies, the energy entropy and normalized energy of IMFs are used as a feature vector to classify the weld quality combined with extreme learning machine (ELM). The result shows that CEEMDAN and ELM can be used to identify the weld quality types of GMAW accurately.