Research On Penetration State Identification Of Pipeline P-GMAW Weld Based On Arc Sound Signa

The increasing scale of long-distance pipeline construction has promoted the development of fully automated welding technology,which effectively improves welding productivity.Compared with manual welding,real-time monitoring of welding quality in the automatic welding process is more important and difficult.How to real-time,fast and accurate access to dynamic process information and real-time quality control is the basic premise to solve the problem of welding quality,the weld penetration state is one of the key indicators to evaluate the quality of welding.Arc sound signal as one of the welding process related signals,is very important for the quality control of the welding process.This paper focuses on the arc sound of pipeline pulsed gas metal arc welding(P-GMAW)to investigate the link between the arc sound signal and the weld penetration state,and to achieve the identification of the weld penetration state.First,the pipeline P-GMAW welding test arc sound signal acquisition system was built,including the arc sound signal acquisition system hardware platform and software system.In order to identify the weld penetration state,design tests to obtain different welding speed,welding current,arc length correction factor under the PGMAW arc acoustic signal library,to explore the relationship between the arc acoustic signal and the change in weld penetration state under different welding parameters.Secondly,in order to investigate the correlation between the arc acoustic signal and the weld penetration state,the arc acoustic signal is pre-processed by applying modern signal processing methods such as framing and short-time windowing.The arc acoustic signal is analyzed in three dimensions: time domain,frequency domain,and time-frequency domain.(1)The time domain characteristics of the arc sound signal are studied and found that the same feature has different ability to characterize different penetration states;the standard deviation and root-mean-square characteristics of the arc sound can reflect the change of the weld penetration state.(2)Using the mean periodogram power spectrum estimation method to process the arc sound signal in frequency domain,it was found that the power spectrum density energy distribution of the arc sound signal changed with the change of weld penetration state and showed certain regularity.(3)The arc sound signal was analyzed in the time-frequency domain to find the wavelet packet energy characteristics of the arc sound signal in different penetration states,and it was found that the wavelet packet relative energy in the frequency range of 1.3-3.2 k Hz can better characterize the P-GMAW welding penetration state.The above study shows that the arc acoustic signal can reflect the change of weld penetration state from several characteristic dimensions.Finally,a four-dimensional feature vector was constructed based on the standard deviation,root mean square,and wavelet packet band energy features that better characterize the weld penetration state,and the data were processed using the maximum-minimum normalization method.The genetic algorithm was applied to support vector regression for parameter optimization,and the P-GMAW welding arc acoustic signal feature vector was used as input and three weld penetration states were used as output to establish the support vector machine(SVM)penetration state identification model.Through experimental validation,the model was shown to have good generalization performance.For the actual use of arc acoustic signal for pipeline P-GMAW welding melt-through state identification to do a fundamental study.