电工技术学报  2024, Vol. 39 Issue (11): 3411-3421    DOI: 10.19595/j.cnki.1000-6753.tces.230660
电力系统与综合能源 |
基于业务性能偏差感知的电力通信网路由优化策略
陈亚鹏, 杨阳, 舒乙凌, 谢文正, 周振宇
新能源电力系统全国重点实验室(华北电力大学) 北京 102206
Service Performance Deviation Awareness-Based Power Communication Network Routing Optimization Strategy
Chen Yapeng, Yang Yang, Shu Yiling, Xie Wenzheng, Zhou Zhenyu
State Key Laboratory of Alternate Electrical Power System with Renewable Energy Sources North China Electric Power University Beijing 102206 China
全文: PDF (1580 KB)   HTML
输出: BibTeX | EndNote (RIS)      
摘要 

针对新兴电力业务对时延、可靠性的严苛要求,提出一种基于业务性能偏差感知的电力通信网路由优化策略。在建立“能量-信息”耦合网络模型基础上,量化分析电力业务转发时延与可靠性约束,设置与能量层、信息层重要度相关的业务效用最大化问题,利用虚拟队列积压感知多跳长时路由优化中的业务性能偏差,基于引入记忆空间的改进SARSA(λ)算法,实现信息不确定场景下的电力通信网路由优化。仿真结果表明,所提算法可有效提升业务效用,在转发时延、丢包率方面性能更优,且可通过权重系数调节适应不同场景的差异化业务需求。

服务
把本文推荐给朋友
加入我的书架
加入引用管理器
E-mail Alert
RSS
作者相关文章
陈亚鹏
杨阳
舒乙凌
谢文正
周振宇
关键词 电力通信网“能量-信息”耦合网络路由优化业务性能偏差感知强化学习    
Abstract

In the context of new power system construction, the traditional power grid with the main purpose of energy transmission is developing towards the trend of "energy-information" coupling network. Meanwhile, the development of emerging power services such as "source-grid-load-storage" collaborative interaction also puts forward higher demands for the service bearing capacity of power communication network. Due to the multi-timescale energy regulation requirements in the new power system, the bottleneck of network bandwidth and deterioration of service performance caused by frequent information exchange are becoming increasingly prominent. In response to these issues, this paper proposes a service performance deviation awareness-based power communication network routing optimization strategy, which utilizes advanced artificial intelligence methods to achieve deterministic service demand guarantee for power services.
Firstly, an "energy-information" coupling network model is established. The emerging "source-grid-load-storage" collaborative interaction service leads to frequent information exchange in the new power system, where power communication network is one of the three pillars supporting the safe and stable operation of the power system. Therefore, the reliable transmission of information plays a more significant role in the rational allocation of energy. Furthermore, considering the current situation of power communication system construction in China, a power communication network model is established based on software defined network architecture, in which the end-to-end forwarding delay of services is analyzed. Due to the complex characteristics of network topology and forwarding conflicts caused by multiple services concurrent access, a multi hop and long-term power service reliability constraint model is given. Then, this paper set the network utility as the amount of successfully forwarded service data related to the grid node importance and service information priority, and proposes an optimization problem to maximize global network utility through routing selection strategy adjustment.
On account of the difficulty in predicting information about future route nodes, the multi hop and long-term power service reliability constraints are unrealizable to guarantee in single hop routing optimization. So virtual queues are introduced to achieve deviation perception between current services performance and the constraints, thereby ensuring the satisfaction of relevant reliability requirements. Taking the uncertainty of time-varying network and service information into consideration, this paper uses reinforcement learning algorithm to realize the autonomous learning optimization of packet routing optimization strategy for multiple power services in the network. Aiming at the problem of insufficient convergence in traditional single hop optimization algorithms, the improved SARSA(λ) with memory space is adopted for routing optimization. Along with continuous learning, early failed learning strategies will gradually be forgotten to improve algorithm convergence.
The simulation results show that compared with traditional routing optimization algorithms based on Q-learning and SARSA, the proposed algorithm performs better in terms of forwarding delay and packet loss rate. Specifically, the service utility has been improved by 21.01% and 15.92%, the forwarding delay has been reduced by 11.43% and 7.14%, and the packet loss rate has been reduced by 35.32% and 19.66%. Also, the weight coefficient can be adjusted to adapt to the differentiated service demands of different scenarios.

Key wordsPower communication network    "energy-information"    coupling network    routing optimization    service performance deviation awareness    reinforcement learning   
收稿日期: 2023-05-15     
PACS: TM73  
基金资助:

国家电网有限公司总部管理科技资助项目(52094021N010(5400-202199534A-0-5-ZN))

通讯作者: 周振宇 男,1983年生,教授,博士生导师,研究方向为智能电网通信网络与新技术、电力物联网与现代传感技术、能源互联网信息通信技术。E-mail:zhenyu_zhou@ncepu.edu.cn   
作者简介: 陈亚鹏 男,1997年生,博士研究生,研究方向为电力信息物理融合系统、电力通信网资源分配优化等。E-mail:yapeng_chen@ncepu.edu.cn
引用本文:   
陈亚鹏, 杨阳, 舒乙凌, 谢文正, 周振宇. 基于业务性能偏差感知的电力通信网路由优化策略[J]. 电工技术学报, 2024, 39(11): 3411-3421. Chen Yapeng, Yang Yang, Shu Yiling, Xie Wenzheng, Zhou Zhenyu. Service Performance Deviation Awareness-Based Power Communication Network Routing Optimization Strategy. Transactions of China Electrotechnical Society, 2024, 39(11): 3411-3421.
链接本文:  
https://dgjsxb.ces-transaction.com/CN/10.19595/j.cnki.1000-6753.tces.230660          https://dgjsxb.ces-transaction.com/CN/Y2024/V39/I11/3411