Research On UAV Visual Positioning Algorithm In Autonomous Inspection Of Photovoltaic Farm

With the development of the photovoltaic industry and the cumulative operation time of photovoltaic equipment,the pressure of operation and maintenance in photovoltaic farms has gradually increased.The traditional manual inspection method is more and more difficult to meet the actual needs.The use of UAV for aerial photography in inspection of photovoltaic farms has become the trend of future development.Computer image processing technology is applied to automate the processing and analysis of aerial imagery of photovoltaic facilities.In order to solve the two positioning problems of UAV in the autonomous inspection process of photovoltaic farms,a strategy of pose estimation and correction by machine vision algorithm is proposed in this thesis.Related visual positioning techniques and methods are studied deeply in-depth as well.The main work and achievements of this thesis are as follows.1)In the autonomous inspection of photovoltaic farms,it is difficult for a UAV to collect high-quality images according to a pre-planned path due to the error of its own positioning system,and the angular difference between the airborne camera and the photovoltaic module may cause perspective distortion.In view of the above-mentioned problem of the relative positional relationship between the UAV and the photovoltaic module,this thesis proposes a method to calculate and correct the relative position of the UAV and PV modules by using machine vision.Firstly,the vertices of the PV module is identified,and then the camera projection model is used to calculate the relative position and adjustment of the UAV.After the UAV is corrected to the corresponding position,it is possible to acquire an image in which the photovoltaic module is facing the camera and centered.For the recognition of PV modules,this thesis uses image segmentation,morphology processing,edge detection,contour extraction and screening to accurately identify all modules.2)In the absence of PV modules'installation information,or the deviation from the planned path to an unknown area,it is necessary to identify environment and estimate the UAV's position to provide a basis for the next path planning.For this problem,this thesis proposes to use the similiar features between the overlapping regions ot aerial images to identify,match,splicing and merging images,and draw local maps in order to achieve the positioning of the UAV.By this method,the UAV can estimate the displacement of the two images containing the overlapping regions themselves.Finally,this thesis summarizes the work carried out and looks forward to the future research to be carried out.