Penetration Prediction And Penetration Quality Control Of Asymmetric Fillet Welding For Medium And Thick Plates

There are many fillet welds in ship,aircraft and engineering machinery.The welding process usually needs three procedures: backing,filling and covering.The backing welding as a key link directly affects the joint quality.In order to ensure the good penetration forming of the fillet weld of the medium and thick plate,the method of single-sided V-groove is often used.However,the following problem is that the single-sided groove makes the significant structure difference between the two sides of the weld root,resulting in serious heat conduction difference,which leads to the situation that the penetration forming is not easy to be controlled stably.At present,this kind of structure generally adopts manual welding.Therefore,the realization of quality control of asymmetric fillet automatic welding of medium and thick plate is conducive to improving the product welding quality and production efficiency,and has practical engineering application and scientific research value.In the previous research,it was found that the size of heat input and the distribution of heat energy has an important influence on the forming of asymmetric fillet weld.Therefore,this paper considered the penetration quality control of asymmetric fillet weld of medium and thick plate from two aspects.The main adjustment of welding current realized the control of penetration depth,and the main adjustment of working angle of torch realized the control of penetration center deflection.Due to the penetration morphology cannot be directly measured during welding,the research focuses on the analysis of the rule of penet ration forming,the exploration of the relationship between sensing information and penetration characteristics,and the establishment of a model for penetration control.Therefore,this following contents are studied in this paper.In the study of the relationship between the heat energy distribution and the deflection of the penetration center,the passive vision method is used.According to the characteristics of arc light spectrum of pulsed TIG welding,a vision sensor system is designed to realize the image collection of weld pool under the condition of high reflection.According to the gray characteristics of the image,a variety of filtering,image enhancement and edge detection methods are tried to get the appropriate image processing algorithm,and high-quality molten pool edge is extracted for feature extraction.The relationship model between the deviation of torch and molten pool and the deflection of penetration center is established as the basis for adjusting the working angle of torch in the later stage.In the study of the relationship model between welding parameters and penetration forming,based on the correlation analysis of the previous experimental data,a BP neural network penetration forming prediction model optimized by the Mind Evolution Algorithm is established.The selection of the model input integrates the welding parameters and the real-time weld pool width,and the output can reflect the penetration depth of the blunt edge and the leg length on both sides.The results show that the model has a high prediction accuracy and stability.The prediction model can be used to guide the adjustment of heat input,and can be used in the subsequent research of penetration quality control.Finally,the penetration quality control system based on PC is built,and its function is realized according to the idea of zone control.The system uses data exchange controller and data acquisition card to collect and adjust welding current.The working angle of torch is controlled by stepping motor.In addition,a monitor interface is developed to realize the start and stop of control process,image acquisition and display.The experiment shows that the control system can adjust the welding parameters well according to the model,and basically achieve the penetration quality control of asymmetric fillet root welding of medium and thick plates.