基于Q‒learning的变电站无线传感器网络路由算法

Q-learning based routing algorithm for substation wireless sensor networks

摘要:

电力系统中的无线传感器网络(WSN)可以对工作中设备的状态和环境数据进行实时感知采集，是一种推动智能电网发展的重要技术。针对变电站场景中WSN的网络存活时间、传输时延、传输丢包率上的特殊要求，提出了一种基于强化学习的WSN路由方案。将数据包在WSN的发送过程抽象为一个马尔科夫决策过程(MDP)，根据优化目标合理设置奖励，并给出了基于Q-learning的最优路由求解方法。仿真结果与数值分析表明，所提方案在网络存活时间、传输时延、丢包率等方面的性能均优于基准方案。

Abstract:

Wireless Sensor Network(WSN) in the power system can sense and collect the status of the working equipment and environmental data in real time, which is an important technology to promote the development of smart grid. Aiming at the special requirements of network survival time, transmission delay, and transmission packet loss rate of WSN in substation scenarios, a WSN routing scheme based on reinforcement learning is proposed. The sending process of packets in WSN is abstracted as a Markov Decision Process(MDP), the rewards are reasonably set according to the optimization objective, and the optimal routing solution method based on Q-learning is given. Simulation results and numerical analysis show that the proposed scheme outperforms the benchmark scheme in terms of network survival time, transmission delay, and packet loss rate.