基于改进Q学习算法的低压电力线通信组网及维护方法  

Low-voltage Power Line Communication Networking and Maintenance Based on Improved Q Learning Algorithm

作　　者：崔莹 刘晓胜[1] 徐殿国[1] CUI Ying;LIU Xiaosheng;XU Dianguo(School of Electrical Engineering and Automation,Harbin Institute of Technology,Harbin 150001,China)

机构地区：[1]哈尔滨工业大学电气工程及自动化学院

出　　处：《电力系统自动化》2019年第24期111-118,共8页Automation of Electric Power Systems

基　　金：国家自然科学基金资助项目（51677034）~~

摘　　要：为提高组网稳定性,选取合适的低压电力线通信(LVPLC)拓扑控制方法至关重要。针对现阶段组网方法不具备自学习能力使得对动态变化的拓扑反应能力相对滞后导致网络不稳定的问题,提出一种适用于LVPLC局域网多约束的改进Q学习算法。该算法基于绑定载波侦听多址接入协议,将非对称信道组网系统建模为离散Markov决策过程。通过与未知环境的不断交互,关联注册节点信息,建立路由表,经周期性地在线学习训练,节点选择较优的转发方向,优化以网关为树根的簇树;周期性地轮换代理,维护并更新骨干簇树网的逻辑拓扑,延长网络生命周期,保证组网的稳定性。仿真结果验证了该算法的有效性与泛化能力。In order to improve the stability of the network, it is crucial to select a suitable low-voltage power line communication(LVPLC) topology control method. To solve the existing problems of low network stability caused by the relatively delayed topological response to dynamic changes in the existing networking method without self-learning ability, this paper proposes an improved Q learning algorithm for LVPLC local area network with multiple constraints. Based on the bind carrier sense multiple access(CSMA) protocol, the asymmetric channel network system is modeled as a discrete Markov decision process. Network stability can be improved through the continuous interaction with the unknown environment, the association with registered node information, the generation of routing tables as well as the periodical online training for optimizing the cluster tree rooted at the gateway. But network stability can be improved through rotating the stations periodically, maintaining and updating the logical topology of the backbone cluster tree, as well as extending the network life cycle. Simulation results verify its effectiveness and generalization ability.

关 键 词：能源互联网 低压电力线载波通信 接入控制 IEEE 1901标准 改进Q学习算法 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]