Study On Laser Brazing Quality Online Monitoring Technology

At present,laser welding technology was gradually applied in industrial automation.With the continuous improvement of welding quality requirements,laser welding on-line monitoring technology was the research hotspot for scholars and technicians all over the world.Researches were shown that the plasma,the infrared heat radiation,the visible light and other optical signals can reflect the welding quality during the laser welding process.The optical signals can be collected by the photoelectric sensors in real time,and the quality diagnosis was made to realize intel igent production.However,the current application of online monitoring technology was poor and still was required for further research.Based on the laser brazing technology,the on-line monitoring technology was researched and the experimental data acquisition was made on the self-built laser brazing on-line monitoring system platform.The main research details were to explore the relationship between signal feature parameters and welding quality,and the welding quality diagnosis was based on classification model algorithm.That was beneficial to the on-line monitoring welding technical application.The wavelet transform and empirical mode decomposition method were used to denoise the signals,and the changing rule between the filtered signals and process parameters was analyzed.It was found that empirical mode decomposition was better filtering effect on instantaneous and non-stationary signals.The processed signal waveform was simple and clear.The rough fluctuations and sharp peaks were filtered out.The root mean square(RMS)value of the signals gradually increased with the increase of the laser power.At the same time,the frequency band of the signals gradually increased,and the frequency energy value also gradually increased.As the welding speed increased,the RMS value of signals gradually decreased,and the signals band and its energy value gradually decreased.The wire feed speed increased from 2.0m/min to 3.5m/min,the RMS value gradually increased,and the signals band and its energy value increased accordingly.When the beam diameter increased from 1.7mm to 4.3mm,the RMS value and frequency were not changed much.The fluctuation of the signals was the same tendency with the molten solder area.The linear correlation coefficient was 0.5541.It showed that the two was the linear correlation.The center temperature of the molten solder maintained a stable fluctuation with the signal fluctuations.The features of the signals were extracted to analyze the correlation between the characteristics of the signals and the welding seam.The time-domain characteristic parameters and frequency-domain characteristic parameters were used to specifically describe the change characteristics of different defects,and thus the laser braze weld defects were distinguished.The macroscopic appearance of laser brazed welds was observed obviously.At the same time,the characteristics of their cross-sectional topography and tensile properties were different.Among which the good weld had the best tensile properties.Hole defects and unfil ed defects had the worst tensile properties.There was a correlation between the signals characteristics and the weld quality.The signal feature parameters corresponding to the four weld profiles were highly discriminating and distinguishable.The laser braze weld defects can be diagnosed by the characteristic parameter values.The principal component analysis method was used to optimize the feature parameters,so as to obtain more simplified comprehensive parameters,thereby reducing information overlap between feature parameters to improve the effectiveness of training samples and the recognition accuracy.Based on principal component analysis(PCA)and support vector machine(SVM)models,the characteristic parameters of signals were used to intel igently identify the quality of laser braze welds.The accuracy rate was 96.6%.Based on the idea of a rule expert system,the quality of laser braze welding was determined.And that was used in the data acquisition module of Labview,it can carry out synchronous diagnosis in the process of real-time monitoring of laser brazing,to realize the real-time diagnosis function.