Research On Laser Welding Penetration Monitoring And Control System Using Real-time Vision

In the laser welding process,the weld penetration state is an important indicator to evaluate the quality of welding,moderate penetration in the ideal welding penetration state.Due to the instability of external environmental factors,it is difficult to obtain a uniform weld penetration through a single welding process parameters,the use of sensing means and control technology to monitor and control the welding process,can effectively guarantee the weld penetration state.The dynamic behavior of the features in the laser welding process can largely reflect the weld penetration state,as far as the current sensing means,visual sensing techno logy has the advantage of rich information.Therefore,this paper builds a coaxial visual acquisition system for keyhole and melt pool in the laser welding process,and uses image processing algorithms to process the acquired information to obtain the corresponding feature information.The correlation between the obtained feature information,laser process parameters and melt-through is analyzed,and finally a fuzzy PID controller with feature information as input and welding process parameters as output is designed to carry out an example test on the control system with a mutation-type variable-section plate as the research object.First,a laser welding system platform based on coaxial vision monitoring was built,which includes a laser welding unit,coaxial vision sensing unit,motion execution unit and information processing unit,using coaxial vision to monitor the dynamic behavior of the keyhole and melt pool to obtain dynamic information.Among them,the vision sensing unit adopts active vision sensing technology and passive vision sensing technology respectively to collect images during the welding process and to test different side axis protective gas angles,which can obtain stable keyhole images within a certain range of gas blowing angles.Next,the area of the penetration hole and the frequency of the occurrence of the penetration hole were extracted using traditional image processing algorithms for the images acquired using passive visual sensing technology.For the image acquired using active visual sensing technique,the introduction of laser light source interfered with the melt pool edge information in the image,which was not conducive to the extraction of melt pool features.To address this problem,the traditional image processing algorithm and the U-Net model are used to segment the melt pool for further extraction of melt pool features,and the comparison results show that the deep learning approach is more adaptable.Based on the acquisition of image feature information,the correlation analysis of each feature,laser process parameters and laser welding melt-through state,to determine the penetration hole frequency,welding speed,laser power and laser welding melt-through state has a strong correlation.Since the actual processing,the general use of high-speed variable power welding process mode to improve welding efficiency,then the penetration hole frequency as the input,the laser power as the output,the design of the PID controller and fuzzy PID controller.To obtain the simulation effect on the welding process for system identification,build a simulation model of laser welding penetration control,compare the simulation effect between PID control and fuzzy PID control,the results show that the control strategy for laser welding penetration,fuzzy PID control is more effective.Finally,the laser welding melt-through control system is developed to build a software control system with real-time vision capabilities.First,according to the functional requirements of the system analysis,then selected the software framework,and finally developed a set of laser welding penetration control software.The experimental object is a mutation-type variable-section plate,and a constant processing process is compared with the use of the melt-through control system Experimental results show that the addition of the controller welding results are better,can obtain the basic consistent melt width,melt-through uniform welding quality.