Research On The Key Technology Of Electric Power Inspection Multi-rotor UAV Obstacle Avoidance System

With the development of relevant technology of multi-rotor uav and its rapid commercial application,Chinese power enterprises pay special attention to and try their best to promote the application of multi-rotor uav in power inspection,hoping to explore a new,efficient and safe inspection operation mode of "mainly uav,assisted by labor".However,in the actual inspection work,the uav is currently operated by the pilot,which relies heavily on the pilot’s experience,so the safety is difficult to be guaranteed.The specific problems are as follows: first,the human eye cannot effectively identify the exact location of the high-voltage tower and transmission line,which is easy to lead to collision accidents;Second,it is difficult to effectively avoid the invisible obstacles on the map such as trees and houses when manually operating the uav.Third,the uav’s flight attitude control performance is not stable,and the flight attitude has a large response to external disturbance.These problems seriously hinder the development of inspection technology based on multi-rotor uav in power inspection industry.Therefore,research on how to ensure the safety of uav in flight operation is one of the major problems to be solved urgently in the current industry.Aiming at the above problems,this paper carries out research on key technologies of obstacle avoidance system of multi-rotor uav for power inspection.The main research contents include the following four aspects:1.An obstacle construction method based on the improved Oriented Bounding Box(OBB)algorithm is proposed.Aiming at the problem that the autonomous flight of power inspection multi-rotor uav must keep a safe distance from the high-voltage tower and power transmission line,this paper proposes to surround the high-voltage tower and power transmission line into regular geometry to assist the route planning before the autonomous flight of uav.In this paper,the improved OBB algorithm makes the compactness of the bounding box better than the AABB algorithm and the traditional OBB algorithm,and shortens the running time of the traditional OBB algorithm by more than 90%.2.Study obstacle detection method based on binocular vision,set up corresponding hardware experiment platform,complete the stereo calibration of binocular camera and three-dimensional correction,based on SIFT(Scale invariant feature-the transform,SIFT)algorithm to realize the binocular vision stereo matching,and the application in electric power inspection scenario realized based on parallax figure around obstacles contour detection,the results can satisfy the application requirements.3.A four-rotor uav attitude control method based on Generalized dynamic fuzzy neural network(gd-fnn)was studied,and a four-rotor uav attitude control system based on gd-fnn was designed.The stability of the system was proved by Laypunov theorem.By comparing with the control performance of traditional PID control in roll Angle,pitch Angle,yaw Angle and x,y and z position,it is proved that the overshot of this system is smaller and the response is faster.Moreover,by adding noise to simulate external interference factors of uav,it is proved that the attitude control effect of this system is more stable.4.A double obstacle avoidance method based on the improved OBB algorithm and binocular vision obstacle detection technology is proposed,and the corresponding experimental scheme is designed.