Research On The Method Of Measuring The Groove Width Of Ultra-narrow Gap Welding Based On Vision

Ultra-narrow gap welding with constrained arc by flux flake is a kind of singlepass multi-layer welding technology for thick and extra-thick plates,which has the advantages of higher welding efficiency and lower heat input.However,the constrained arc is very sensitive to the change of groove width.If the welding parameters do not match the groove width,the arc will lose its self-adjusting function and lead to welding failure.Therefore,it is very important to measure the width of each groove accurately.Because the groove of ultra-narrow gap welding is deep,narrow and changes with different welding passes,the conventional measurement method is difficult to apply.To solve this problem,a soft sensing method based on vision and welding thermal deformation model is proposed in this thesis.The main work and innovation are summarized as follows.1)considering the groove characteristics of ultra-narrow gap thick plate welding and the geometric characteristics of weldment,through the analysis of the advantages and disadvantages of various visual measurement systems,a laser visual measurement system with high sensitivity and strong anti-interference ability is designed and built.According to the characteristics of the measured object,a calibration method named coplanar linearity calibration is designed.The performance of the calibration method is studied by the control variable method.The experimental results show that when the number of feature points is more than 10,the accuracy and stability of the calibration algorithm are high whether the number of feature points,the measurement length or the attitude of the calibration plate are changed.2)In order to overcome the uneven brightness distribution caused by the laser line projected on the surface of the weldment and the difficulty in determining the binarization threshold,a method of top-hat transformation combined with Otsu is proposed to equalize the gray level of the laser line and separate it from the background effectively.What's more,to obtain the centerline of single-pixel laser stripe and identify the feature points accurately,a method combining skeleton refinement and least square is proposed.3)For Q235 thick steel plate weldments with different groove widths,the upper width of each pass is measured by the vision system.According to the relationship between the slope angle deformation and the accumulated height of the welding layer,the current groove width model is established.Combining the model with the measurement results of the upper groove width,the width of the lower groove is calculated.The maximum measurement error is less than 0.2mm,which can meet the requirements of ultra-narrow welding process parameters setting on the measurement accuracy of groove width parameters.It can also fully demonstrate the accuracy of the model established for ultra-narrow gap welding object and the effectiveness of the visual measurement system.4)It is of great significance to know the welding layer height accurately for setting parameters before welding,welding process monitoring,quality evaluation after welding and soft measurement of groove width in ultra-narrow gap welding.Aiming at the problem that the welding layer height is difficult to be measured directly,a prediction method based on Light GBM is proposed.Combined with the mechanism and data fusion analysis method to determine the characteristics,and on this basis to establish a model for prediction.The experimental results show that this method can accurately predict the welding layer height,and its MAE is 0.114 mm.This method can be extended to ultra-narrow gap welding of rail and other extra thick plates to solve the problem that it is difficult to directly measure the width of weld bead groove in ultra narrow-gap welding.It can sense the change of groove width in time to guide the adjustment of process parameters and realize the goal of the intelligent welding process.