Improving UAV-Based Monitoring of Solar Power Plants Using Coverage Path Planning Model with Adaptive Learning Algorithm

Solar power plants, as one of the key renewable energy sources, require advanced solutions for effective monitoring and enhanced efficiency. This paper introduces a novel Coverage Path Planning (CPP-ALA) algorithm for monitoring solar power plants using UAVs, designed to improve energy efficiency and maintenance operations by employing real-time image processing, adaptive learning, and deep learning techniques. The proposed algorithm utilizes dynamic adaptive learning rates, a batch size of 16, and the Adam optimizer, combined with ReLU and Sigmoid activation functions, for the detection of defects in solar panels. Experiments conducted on UAV-acquired aerial images of solar power plants evaluated the performance of CPP-ALA against three other methods: semantic segmentation-based coverage planning, multi-agent reinforcement learning (MARL), and wavefront coverage planning. The results demonstrated that CPP-ALA achieved 97.8 % defect detection accuracy, a processing time of 1.8 seconds per image, and a localization accuracy of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\pm 5$</tex> pixels. Moreover, the algorithm exhibited a 30% reduction in computational complexity, enabling real-time implementation. This approach establishes a new standard in UAV coverage path planning and defect detection, providing an efficient solution for solar power plant inspection and supporting renewable energy applications.