Research On Automatic Tracking System Of Film Wall Welding Seam Based On Laser Vision Sensing

Aiming at a large number of heterosexual film wall welds in the boiler industry,this article develops a 5-degree-of-freedom welding robot based on monocular linear structured light based on the gantry platform.This robot has a wide working range and strong adaptability,and is particularly suitable for Membrane wall welding automation operation in the boiler industry.The welding robot uses external sensors such as monocular linear structured light sensors.It has the characteristics of stable operation,long working time,and strong anti-interference ability.Therefore,monocular linear structured light is used in The actual welding automation is of great significance.This paper focuses on the construction of hardware systems,software system design,image processing and recognition of membrane wall welds,weld deviation identification,robot trajectory planning and controller design.Firstly,the composition of the robot system is introduced.The hardware part is composed of a gantry rail-type mobile platform,a welding torch,an arc guide mechanism,a monocular linear structured light sensor,and a robot control box.The monocular linear structured light sensor is designed and processed.It is reasonably installed on the gantry through the connection board,and the CCD camera is used to collect the image emitted by the laser on the film wall weld to prepare for the subsequent weld image processing.The software part is from the OpenCV open source library and some boards in the control system.The card comes with SDK and MFC,etc.,and is developed in the Window 7+Visual Studio 2013 environment.Secondly,the camera in the sensor is calibrated,a mathematical model of the vision system is established,and the welding seam image processing and welding seam feature point extraction algorithm are designed,which mainly include gray-scale open operation,top hat transformation,connected domain marker deletion,The gray level frequency determines the laser fringe area,the morphological operator refinement processing,the skeleton extraction algorithm and the dynamic ROI weld feature point extraction method,and establishes a mathematical model for weld deviation recognition,and tracks the film wall weld seam.Planning,based on the cubic uniform B-sample interpolation surface fitting method and fuzzy control algorithm,a controller for automatic tracking system of membrane wall welds was designed.Finally,welding tests were conducted on the membrane wall space welds in the boiler industry,which were divided into flat welds,uphill welds,downhill welds,and the front and back of the membrane wall,and experimental research was conducted on them.Welding seam tracking accuracy and simple process analysis were done.The experimental results show that the design of the hardware part in this paper is reasonable,and the designed welding image processing and tracking control algorithm is robust,reliable,accurate,and accurate.