Research On Control System Of Detection Arm For Inspection Robot On Transmission Line

Overhead line is an important part of power system.Exposed to outdoor environment for a long time,overhead line and kinds of fittings on lines are prone to damage,so it is necessary to inspect and repair the overhead lines and fittings.Using overhead line robot instead of manual inspection can reduce labor intensity and protect personal safety.Due to the low intelligent level of overhead line robot and less consideration of the robot swinging with the wire,the path planning technology of the detection arm of overhead line robot,the position compensation technology for the small angle swing of the robot platform and the swing prediction technology for the large angle swing of the robot platform are studied.Firstly,the control system of overhead line robot detection arm based on ROS(Robot Operating System)is designed and implemented.Secondly,the forward kinematics and inverse kinematics of the detection arm are analyzed,and the path planning algorithm for the detection arm to complete the detection task is studied and implemented,mainly including Astar algorithm and Informed RRTstar(Rapidly-Exploring Random Tree)algorithm.Then,the swing posture of the overhead line robot on the wire is analyzed,and the position compensation model of feedforward control based on kinematic analysis for the small angle swing of the overhead line robot platform and the large angle swing prediction model based on real-time variance statistics for the large angle swing are designed,which predict the emergence of large angle swing by calculating the variance change in the window.Finally,the algorithms and models are verified by simulation and real experiments.The comparative experiments between Astar algorithm and Informed RRTstar algorithm show that Informed RRTstar algorithm has higher success rate and less time-consuming,but the execution time of the planned path is longer.Considering that the total time-consuming and the path search time of Informed RRTstar algorithm is less when there are wire obstacles,Informed RRTstar algorithm can complete the path planning task of detection arm for detection better.The position compensation module of feedforward control based on kinematics analysis can reduce the offset of the actuator at the end of the detection arm due to small angle swing,and avoid the collision with the line.Compared with the empirical mode decomposition and support vector regression models,which based on time series prediction,the large angle swing prediction model based on real-time variance statistics has higher recall and F-measure,but lower accuracy.Because recall is the main evaluation index of the swing prediction model of overhead line robot,the large angle swing prediction model based on real-time variance statistics is more suitable for the large angle swing prediction of overhead line robot.