基于深度强化学习的综合能源业务通道优化机制

为了保障综合能源系统的稳定运行，承载综合能源业务的通信网络需要具备高可靠、低风险等特征.依据综合能源业务的通道要求，提出了一种深度强化学习的算法，旨在对大规模综合能源业务在承载的电力通信网上寻找到整体最优的路径.该方法以整体时延和网络负载均衡度为目标，对网络拓扑进行训练，并保存模型，然后通过迭代学习获取最优的结果.仿真结果表明，该方法找到的路径既可以保证整体时延较短，又可以保证网络的整体负载均衡.同时，在网络规模很大、业务数量很多的情况下，深度强化学习算法可有效提高计算效率.

A Integrated Energy Service Channel Optimization Mechanism Based on Deep Reinforcement Learning

In order to ensure the stable operation of the integrated energy system, the integrated energy service needs to have high reliability and low risk when being carried by the communication network. According to the channel requirements of the integrated energy service, an algorithm of deep reinforcement learning is proposed, aiming to find the overall optimal path for the large-scale integrated energy service on the carried power communication network. The method that aims at the overall delay and network load balance, trains the network topology and saves the model, and then obtains the optimal result through iterative learning. The simulation results show that the routing found by this method can ensure the overall delay is short and guarantee the overall load balance of the network. At the same time, for scenarios with a large network size and a large number of services, the deep reinforcement learning algorithm can effectively improve the computational efficiency.