- This system is a comprehensive software-hardware integrated platform for inspection training of photovoltaic (PV) power stations. Deployable in an 8m×5m indoor space, it is specially designed for PV power station inspection simulation training and intelligent recognition algorithm verification in GPS-denied indoor environments. The system consists of three modules: small PV inspection UAVs, PV inspection control and management platform, and PV power station simulation scenario. The small PV inspection UAV is fitted with LiDAR, optical flow positioning, a domestic onboard computer, a three-axis gimbal stabilizer, and an ultra-high-definition visible-light/infrared thermal imaging dual-sensor camera, to perform PV power station inspection flights and data acquisition. The PV inspection control and management platform adopts B/S (Browser/Server) architecture with on-premises deployment, enabling 3D scenario display, real-time flight monitoring, mission management and AI defect recognition algorithm deployment. The PV power station simulation scenario includes PV panel arrays, combiner boxes, inverters and support structures, replicating the real inspection operation environment of PV power stations.
- Equipped with a LiDAR sensing module and an optical flow positioning module, the system achieves high-precision positioning, spatial modeling and real-time obstacle avoidance in GPS-denied indoor environments. It can accurately identify the spatial positions of equipment such as PV panel array boundaries, combiner boxes and inverters, and underpin algorithm verification and training for UAV autonomous navigation, flight path planning, AI defect recognition and other applications.
- The system platform supports 3D scenario display (with import of 3D scenario models of PV power stations), display of planned flight paths and real-time UAV trajectories, access to real-time visible/infrared video streams, and presentation of recognition results. It also enables real-time monitoring of UAV telemetry data (position, motion status, battery level, mission status), inspection mission management (flight path planning, mission scheduling, data management), and remote command control (takeoff/pause/continue/return/landing).
- The system supports synchronous acquisition of visible-light and infrared thermal imaging dual-sensor images and AI defect recognition. It can identify common defects including PV panel hot spots, hidden cracks, breakage, fouling and shading. Inspection data can be automatically recognized and annotated by the configured recognition algorithms, meeting the training demands of artificial intelligence courses such as PV power station defect recognition and component identification.
- The system forms a closed-loop integrated teaching mode of "scenario cognition – simulated flight – defect recognition". Students can repeatedly practice core skills such as flight path planning, dual-sensor data acquisition, defect annotation and inspection report generation in the simulation scenario, making it suitable for practical training of courses including UAV PV inspection technology and new energy operation & maintenance.