Time frequency feature extraction of the arc energy for quality detection of the aluminum alloy double pulse MIG welding

Abstract

In aluminum alloy double pulse metal inert-gas (DPMIG) welding, the outputting arc current waveform is distorted under the influence of various factors such as harmonics and impact loads. The arc distortion of aluminum alloy DPMIG welding affects the arc stability and welding quality. Based on the collected welding current signal, a feature extraction method is proposed for quality detection of the aluminum alloy DPMIG welding. The wavelet method is adopted to eliminate the noise of the welding current signal. The local mean decomposition (LMD) is performed to the welding current signal to obtain a series of Product Function (PF) components with real physical meaning. The Hilbert transformation is subsequently performed to the PF components to obtain the time frequency distribution of welding arc signal energy. The approximate entropy (ApEn) of the time frequency distribution of the welding current signal is calculated to evaluate the arc stability and the welding formation quality. Application of the proposed feature extraction method indicates that the combination of the wavelet and LMD can effectively extract the distortion components of the welding current signal. The time frequency distribution of the PF components of the welding current can clearly reflect the concentration and dispersion of the arc energy. The approximate entropy of the time frequency distribution of the welding current can be quantitively reflect the arc stability and the welding formation quality in the aluminum alloy double pulse MIG welding.