X-ray Inspection Robot And Its Application In Automatic Detection Of Weld On Large Sized Ellipsoidal Head

Ellipsoidal head with large size is the essential part of pressure vessel in numerous industries such as petrochemical industry, nuclear industry and food and pharmaceutical industry. The head is stamped from two different steel plate with random size after welding. For security, the weld must undergo strict X-ray inspection before the head equipped. The current method of the inspection is manual detect carried out by worker, which exposed the worker in high dose X-ray radiation. Meanwhile, the efficiency and accuracy of detection is disfactory. This paper designs an X-ray inspection robot according to the requirement of automated inspection of large sized ellipsoidal head. The main results include:A scheme of X-ray inspection robot based on visual guiding is proposed according to the requirement of automated inspection of large sized ellipsoidal head. The robot has 5 DOFs,and is consisted of robot body, visual system, control system and auxiliary system.A method of weld image extraction and weld localization based on visual measurement is proposed according to the requirement of automated recognition and localization of weld on ellipsoidal head. Combined with robot kinematics, a path planning method based on quintic interpolation is proposed and verified by simulation. This part of the research lays the technical foundation of the robot motion control.For the situation that the robot arm will suffer from severe deformation in working condition which results large detection error, a new method of online compensation for deformation is proposed. This method analysis the deflection of robot arm through FEA method, and fits the function of deformation by RLS method, and proposed iterative compensating strategy for the deformation. Results from the experiment show that the method significantly improves the accuracy of inspection.Finally, an introduction of the realization of the scheme including mechanical structure design, hardware selection, software design, is offered. Experiments prove the robot system is capable of automated inspection of weld on ellipsoidal head.