Design Of Photovoltaic Module Cleaning System Based On Drone

In the photovoltaic power generation system,the deposition of particles such as dust and bird droppings on the surface of the photovoltaic modules will reduce the efficiency of the photovoltaic system that works for a long time,and affect the normal operation of the photovoltaic power station.In severe cases,hot spots will also occur,which will damage the photovoltaic equipment.Aiming at the cleaning problem of photovoltaic modules,this paper studies and designs a cleaning system that uses drones as a carrier and can autonomously perform cleaning tasks for photovoltaic modules.In the cleaning task,the system can accurately identify and locate the stains and cracks on the photovoltaic modules,and achieve the required cleaning effect by changing the water pressure and spraying method.In addition,the system also uses image recognition,track planning,intelligent control and other technical methods to implement autonomous navigation and trajectory optimization functions.The specific research work of this paper is as follows:（1）A set of airborne photovoltaic cleaning water spray device is designed,which can drive the water pump according to the preset control instructions to realize the function of spraying cleaning agent by adjusting different water pressure and spraying methods..Aiming at the influence of the recoil force brought by the water pump during the photovoltaic cleaning process on the stable state of the UAV,the system improved and designed a double closed-loop control algorithm that can meet the requirements of compensating the rotor speed and maintaining the stability of the inclination angle of the UAV during the cleaning process.The flight platform also realizes data interaction with the ground host computer through wireless communication,which further enhances the ground-air synergy of the cleaning system.It can obtain flight data from the host computer and realize long-distance photovoltaic cleaning operations.（2）According to the specific working conditions of the task,the fixed-point control flight method is adopted,and the UAV is better adapted to the photovoltaic field and distributed photovoltaic working environment through track planning.After the operator manually performs a task,the UAV can repeat the cleaning process of the fixed-point photovoltaic device according to the recorded task path,through the previous operating instructions and the real-time tracking function of the UAV flight data.（3）According to the characteristics of photovoltaic modules,YOLOv3 algorithm is selected to realize the regression extraction of photovoltaic panel regions in video images.It can independently complete the work of track planning and flight,photovoltaic panel positioning and cleaning,and returning to flight and adding water,which greatly improves the autonomy of the photovoltaic cleaning system.（4）Use a two-stage inspection algorithm to inspect the surface of photovoltaic modules,perform special cleaning treatment on existing deposits,and issue warnings for cracks and other faulty defects.On the MS COCO data set,the R-FCN small target detection algorithm improved by Faster R-CNN is adopted.With the same accuracy as the former,the detection time is greatly shortened,and the dynamic real-time performance and work efficiency are improved.