Experimental Study On Magnetic Abrasive Finishing Of Bend Pipe Inner Surface Based On Profile Recognition

With the advancement and development of science and technology,my country's equipment manufacturing industry is moving towards intelligence and precision.While pursuing excellent performance,it also has higher requirements for the service life and stability of the equipment.The bend pipe has the characteristics of high space utilization and flexible application,and is widely used in aviation,medical,marine and other pipeline systems.Due to the influence of the forming and manufacturing process,defects such as wrinkles and micro-cracks are prone to occur at the bends.During the long-term operation of the equipment,these pipeline defects become potential failure of the equipment,which seriously affects its operational reliability.Therefore,there are new requirements for the internal surface quality and polishing treatment methods of the bend pipe.Because magnetic abrasive finishing is a flexible grinding process and has the characteristics of small grinding particles that can enter a narrow space for operation,it is currently used in the polishing of the inner surfaces of bend pipe.However,for the generation of magnetic abrasive finishing processing trajectory for the bent pipe,the usual method is to manually collect discrete points to generate the robot trajectory,which method disadvantages are low efficiency,large random errors,and difficult posture adjustment.Based on this,the paper proposes a magnetic abrasive finishing test method for the inner surface of the bend pipe based on profile recognition,which uses the midline of the bend to quickly and accurately generate the grinding trajectory.In this paper,through the analyzes of magnetic abrasive finishing mechanism,a single abrasive particle's force state during the grinding process,and the change in the magnetic field intensity of the grinding area.Determine the key process parameters of the magnetic abrasive finishing on the bend pipe inner surface,and basis on this for calculation of pose in the robot grinding process.The image of the bent pipe is used for extract the midline of the profile,and through the calibration to realized the transformation from the pixel points in the image to the coordinate points in the robot base coordinate system.Use bilateral filtering to remove image noise and retain the edges of the bend pipe.Binarize the filtered image and use the area selection algorithm to retain the bend pipe images that need to be grinding.The morphological thinning method and the m-adjacent point branch removal method are used to generate The centerline of the smooth bend pipe.The midline of the bend is traversed to obtain the pixel coordinates of all points on the midline,which is equidistantly scattered and then converted into data points in the robot base coordinate system.The data points in the base coordinate system are interpolated cubic non-rational B spline to obtain a smooth bend midline function and its first order derivative.According to the magnetic abrasive finishing process requirements of the bend pipe inner surface and the centerline function,the grinding movement pose of the magnetic abrasive finishing device is determined.On this basis,the movement interference check is performed to obtain a suitable grinding gap.Finally,the Win Socket network communication between the computer and the robot controller is used to interact with the data and drive the robot to drive the grinding device to reciprocate the grinding process of the bent pipe.The test results show that the processing accuracy of the bend pipe meets the requirements,and the whole test process can significantly improve the data points collection efficiency of the bend pipe midline,avoiding random errors caused by manual measurement and increasing the consistency and accuracy of data point collection.