Research On The Identification Method Of GMAW Weld Deviation Of Medium And Heavy Plate Based On Arc Acoustic Signal

Medium thickness plates welding has been widely used in ships,heavy machinery and equipment and other fields.The use of automated welding methods can effectively improve production efficiency,but the workpiece is affected by thermal deformation and other factors during the welding process,which will cause deviations in the welding seam,which will lead to welding defects.Therefore,automatic welding seam tracking technology plays a very important role in ensuring welding quality.Arc sound sensing is a non-contact collection method,with low hardware cost,and the collection process is simple and easy.The arc sound signal contains a lot of welding process information,and the arc sound will change when the welding seam is deviated.Therefore,the relationship model between the characteristics of the arc sound signal and the weld deviation can be established to realize the recognition of the weld deviation.The thesis mainly studies the recognition of GMAW weld deviation of medium thickness plates based on arc sound signals.First,the GMAW welding test signal acquisition system was established.The generation mechanism of the arc sound signal is studied,and the results show that the arc sound signal is mainly derived from the change of arc energy.The polynomial least squares method is used to preprocess the original arc sound signal for the de-trend term,and the processed sound signal waveform moves to a symmetrical position about the origin,ensuring the accuracy of the signal.The fast Fourier transform method is used to analyze the arc acoustic signals with different swing periods,and it is found that they have the characteristics of highly similar frequency characteristics.Based on the above analysis and considering the swing period factors,the complete selection of acoustic signal samples is realized.Secondly,in order to study the correlation between arc sound and weld deviation,the characteristics of arc sound signal were analyzed from three dimensions: time domain,frequency domain,and time-frequency domain.(1)Studying the characteristics of arc sound signal waveforms found that: intercepting 75 ms data at the left and right extreme positions to carry out the time domain analysis results are the most stable;the characteristics of arc sound energy and variance on both sides of the weld can reflect the change of weld deviation.(2)Using the Welch power spectrum estimation method to analyze the arc sound signal in the frequency domain,the power spectrum density and energy distribution of the arc sound signal changes accordingly with the change of the weld deviation.(3)Analyze the time-frequency domain characteristics of the arc sound signal,obtain the wavelet packet frequency band energy characteristics of the arc sound signal under different weld deviation conditions,and find that the wavelet packet energy of the seventh and eighth frequency bands shows the best deviation sensitivity Sex and stability.The above research shows that the arc sound signal can reflect the change of the weld deviation from multiple characteristic dimensions,which is closely related to the weld deviation.Finally,a four-dimensional feature vector is constructed based on the energy difference,variance difference and wavelet packet frequency band energy features with high sensitivity to weld deviation,and the maximum and minimum normalization method is used to process the data.Grid search algorithm and genetic algorithm are used to optimize the parameters of support vector regression,and the support vector regression model of grid search optimization with the best weld deviation recognition effect is selected as the final model.And the validity of the welding seam deviation recognition method proposed in this paper is verified by experiments.It provides a new solution for real-time tracking of GMAW welds of medium thickness plates by using arc sound.