Identification Of Droplet Transfer In MIG Welding Based On Arc Acoustic Signal

In the process of gas shielded arc welding,how to monitor and control the quality of welding is one of the most important subjects.In the actual production,the welding quality determines the quality of the final product,and the droplet transfer mode not only determines the stability of the welding wire melting time but also seriously affects the weld forming and weld depth and material consumption,metallurgy,and is closely related to the welding quality.According to the MIG welding arc sound signal in the process of tile sheet was studied,the arc sound signal recognition based on different droplet transfer type.Synchronous acquisition and analysis system to set up the test platform for MIG welding with different droplet transfer mode of arc sound signals and electrical signals,including welding robot system,composite sensing system,the droplet transfer mode of high speed photography system and system software.According to the characteristics of MIG welding method,developed a suitable signal sensor system and arc acoustic sensor system,the image signal can be effectively collected in the process of welding arc sound signals,signals and droplet transfer state.Analysis of the autocorrelation function of the short circuiting transfer mode of arc sound signal and electric signal,and analyze the cross-correlation function of arc sound signal and current signal and the arc sound signal and voltage signal shows that the arc sound and the voltage and current signals with periodic similar,arc energy and arc sound are closely related.The arc sound signal for short circuiting,shotdrops transition and jet transition to the power spectrum analysis,power spectrum waveform can be found by different droplet transfer mode,the frequency distribution of arc sound signal have obvious difference,and has certain regularity,short circuiting process more low frequency components,shotdrops transition and the transition of high frequency jet there are many.The wavelet packet analysis of arc sound for short circuit transfer,droplet transfer and jet transition.Wavelet packet decomposition wavelet basis function selection db14,decomposition layer set to 4.Extracting the characteristic values of the band energy after the 4 layer decomposition of the arc acoustic signal.The distribution of S4,0,S4,2,S4,3 band energy distribution of arc sound signal has obvious difference,which can be used as the characteristic vector to identify the droplet transfer type.From numerical,different droplet transition state of arc sound signal kurtosis system analysis,short circuiting,shotdrops transition and the transition of the Ku jet kurtosis differences,can be used as a feature vector of droplet transfer type.In view of this,the identification of the four dimensional joint feature vector of the droplet transfer mode is completed.Based on the MATLAB software platform,the network model of the droplet transfer pattern recognition for MIG arc welding is designed,and the generalized regression neural network and probabilistic neural network are selected.The results show that the recognition rate of the droplet transfer type of GRNN network is96.6667%,and the recognition rate of the PNN droplet transfer type is.It can effectively identify the type of droplet transfer,and the recognition accuracy is high,which can meet the expected target.