Cooperative scheduling of rural public resources based on multi-agent proximal policy optimisation

To address the inefficiency in allocating public resources within rural areas, this study proposes an innovative collaborative scheduling framework based on multi-agent proximal policy optimisation.This approach employs a decentralised decision-making mechanism, enabling multiple agents to learn cooperative strategies thereby optimising the scheduling of shared water and energy resources among interconnected villages.Experimental results demonstrate that this approach effectively captures demand characteristics across diverse rural regions, coordinating the rational exchange of public resources.Specifically, compared to single-agent deep reinforcement learning algorithms, the proposed solution reduces total operational costs by 1.72%, and enhances renewable energy consumption, delivers superior performance in both economic and environmental benefits.These findings validate the broad prospects of multi-agent reinforcement learning as a practical solution for automated intelligent scheduling of rural public resources.