Research On Visual Recognition And Servo Control Method Of Insulator Cleaning Robot In Substation

Insulators in substations have been in the open air for a long time,and their surfaces are affected by natural factors such as salt spray and climate,and are susceptible to moisture and contamination.As a result,pollution accumulation on insulators’ surfaces will result in flashover discharge accidents,which will cause harm to the power system.The traditional manual power outage cleaning method has great economic loss,and the manual electrification cleaning method has great labor intensity,low operating efficiency and high risk.Therefore,it is necessary to research and design a compact insulator cleaning robot to realize the duty of live cleaning of insulators in the substation.Based on the existing insulator cleaning robots at home and abroad,this paper studies and designs a compact cleaning robot,which is equipped with a dry ice cleaning module to realize multi-directional cleaning of pillar insulators through the dry ice nozzle at the end of the operation of the manipulator arm.In this paper,the method of robot visual recognition and localization and the control method of manipulator are studied.Firstly,according to the operating environment and requirements of the pillar insulator cleaning robot,the design scheme of the key structure of the robot,such as the crawler lifting platform and the manipulator arm operating platform,is proposed.Secondly,the robot vision algorithm is studied.The YOLO-V4 target recognition algorithm is applied to identify and detect the insulator and the connecting flange,and the three-dimensional positioning of the insulator is carried out by combining the bounding box prediction with the camera depth information.In addition,the image processing method is used to detect and extract the details such as the edge and Angle of the insulator fan skirt,so that the subsequent rotary washing mechanism can accurately clean the concave part inside the insulator plate umbrella skirt.Based on the target recognition and location algorithm,the coordinate system of each joint link of the robot arm was established,and the D-H parameters were obtained to carry out the kinematics analysis.The hand-eye servo system of the manipulator was established in the way of eye-in-hand,and the calibration of the manipulator and the visual system was completed.The spatial conversion relationship between the insulator,the camera and the manipulator was established,and the attitude of the target insulator was estimated.Cartesian space and joint space were used to plan the motion of the end of the manipulator,and MATLAB Robotic Stoolbox tool was used to simulate the motion of the end of the manipulator.Secondly,the fuzzy PID controller is constructed by combining the fuzzy control theory with PID control theory to control the joint of the manipulator.Finally completed the development of the robot prototype,build experiment environment to control the robot to carry out the visual experiment,it has been verified by experiment that the robot vision system is to be able to identify location of insulator,each joint movement flexible mechanical arm,the stable control of the desired cleaning effect,the effectiveness of the robotic visual servo control method was verified.