GAT-PPO-Based Joint Routing Optimization for TSN-Enabled VPP Backbone Networks

As virtual power plants continue to scale and time-sensitive services increase, wide-area backbone networks face stricter requirements for low latency, high reliability, and efficient transmission. Existing VPP backbones have difficulty sustaining stable end-to-end quality of service under dynamic mixed traffic and time-varying loads. To address this issue, this paper proposes a joint routing and scheduling optimization method for VPP wide-area backbone networks based on graph attention networks and proximal policy optimization, while incorporating a dynamic priority-sorting mechanism under cyclic queuing and forwarding constraints. The problem is formulated as a Markov decision process, where graph attention is used to extract local topology and link-state features, and proximal policy optimization enables hop-by-hop online routing decisions, thereby improving transmission adaptability in dynamic scenarios. Simulation results show that the proposed method outperforms baseline schemes in average end-to-end latency, packet loss ratio, and effective throughput, while also exhibiting favorable convergence performance.