Research On Automatic Path Planning Of UAV For Wind Turbine Blade Inspection

With the development of UAV and image processing technology,the inspection of wind turbine blade surface damage based on rotor UAV has achieved great development.However,the electromagnetic interference of wind farm will seriously affect the GPS(Global Positioning System)positioning of UAV,and the inspection route planning is mainly based on the experience of operators,so the efficiency is difficult to guarantee.Therefore,in order to improve the inspection efficiency and positioning accuracy,the optimal path planning and visual inertial fusion positioning for UAV automatic inspection of fan blades is studied as follows:(1)Considering that the UAV needs to obtain a clear image set that completely covers the surface of the wind turbine blade when inspecting the wind turbine blade,this dissertation first establishes the wind turbine model,and grid the model,and then plans the candidate track point space corresponding to each grid on the surface of the wind turbine model according to the camera angle of the imaging equipment,the setting of the safe distance and other constraints,Ensure that the candidate track points can capture at least one grid area,and finally complete the initialization of the track points by random sampling.(2)The 3D scene target modeling and optimal trajectory planning design for the inspection of wind turbine blades.Since the planning process of optimal trajectory is equivalent to the process of solving NP hard problem,this work uses iterative re optimization algorithm for 3D scene target modeling and trajectory planning.Based on the sampled track points,RRT* is used to calculate the initial cost between the track points,and lkh-tsp algorithm with Variable k value is used to optimize and re optimize the track line,so as to obtain the current optimal track line;Then,based on the above idear,the shortest track point set is resampled from the candidate track point area constructed according to the constraints,and the optimal track line is replanned according to the resampled track point set.The two steps are carried out alternately until the optimal track line that can pass through all the track points without repeating is obtained.Finally,the simulation is carried out based on ROS in the Ubuntu system,and the optimal path planning of 3D scene inspection of fan blade is completed.(3)UAV patrol location calculation and accurate positioning based on binocular vision slam model.In order to overcome the influence of the outdoor wind field environment light on the visual positioning,this dissertation adopts the method of binocular visual inertia fusion for positioning,and the two cooperate with each other in a tight coupling way.Inertial measurement makes up for the problem of tracking and positioning failure when visual positioning moves fast or rotates purely.Visual slam can achieve accurate positioning in a long period of time,reduce the cumulative error of inertial measurement,and match the point and line features in the visual positioning system,make full use of the wind environment information,and improve the positioning accuracy and robustness.(4)The real-time optimization adjustment of the global track line considering the deflection angle of the fan blade.Due to the uncertainty of the shutdown state of the blades in different wind farms,this dissertation proposes a global trajectory optimization adjustment strategy which can identify the blade deflection angle online.Firstly,the UAV flies to the front of the pre inspection fan cabin to obtain the front image of the fan;Then,edlines algorithm is used to obtain the edge information of the fan blade stop contour,and the least square fitting method is used to clear the image;Finally,the deflection angle of the fan blade relative to the reference state is calculated,and the global track line is optimized in real time according to the track point adjustment formula given in this dissertation,so as to ensure that the planned track line can be used to inspect the fan blade under any shutdown state.(5)Design software and hardware simulation platform.In the ROS environment,the 3D model of the wind turbine blade is constructed to verify the calculation ability of the proposed algorithm for the optimal track line and the accuracy and stability of positioning.At the same time,the hardware platform is built,the vision sensor and inertial sensor are calibrated,the quantitative and qualitative experiments are designed,and the feasibility of adjusting the track line considering the deflection angle of the fan blade is verified.