Objective Aiming at the problems of low efficiency in manual inspection of large-scale photovoltaic (PV) power plants, limited coverage capability of a single unmanned aerial vehicle (UAV), and the difficulty in balancing fairness and energy efficiency in multi-UAV cooperative path planning, this paper aims to propose a cooperative inspection path planning method for UAV swarms based on multi-agent deep reinforcement learning to improve the level of intelligent operation and maintenance.

Method Firstly, the inspection task was modeled as a multi-objective joint optimization problem considering geographic coverage fairness, UAV load fairness and swarm flight energy consumption. Secondly, a hierarchical cooperative decision-making framework based on Multi-Agent Deep Deterministic Policy Gradient (MADDPG) was designed. Adopting the "centralized training and decentralized execution" paradigm, each UAV acted as an independent agent to autonomously decide flight direction and distance in continuous action space. Meanwhile, a low-complexity nearest-assignment strategy was introduced to complete inspection target allocation, reducing decision dimensions.

Result Simulation results show that compared with baseline methods such as random flight and fixed circular trajectories, the proposed method is significantly superior to traditional strategies in terms of coverage completeness, task fairness, and energy consumption control, providing an efficient, intelligent, and scalable UAV swarm inspection solution for large-scale ground-mounted PV power plants.

Conclusion This method can effectively solve the problems of coverage imbalance and load imbalance in large-scale ground-mounted PV power plant inspection, providing an efficient, intelligent, and scalable UAV swarm inspection solution for PV operation and maintenance, with good potential for practical application.