基于图神经网络的电力通信网路由优化算法

Routing Optimization Algorithm for Electric Power Communication Network Based on Graph Neural Network

随着社会发展和用电需求的增多,在电力通信网络中高可靠地传输关键业务数据成为业内关注的重点。在自然灾害、人类社会活动等因素导致网络拓扑改变或者电力业务的通信需求发生变化时,快速建立业务的可靠路由是一个复杂且极具挑战的问题,很难通过传统的计算手段实时求得最优路径组合。通过分析电力通信网络结构和业务传输需求的特点,建立主备路由和单路由的协同优化问题,将图神经网络与深度优先搜索算法相结合,以业务传输的可靠度为优化目标,提出了适应于不同网络场景和业务需求下的高可靠路由方案。仿真结果表明相较传统路由算法,所提算法在时延、容量等工程约束下可获得更高的路由可靠度,并且可以推广到不同结构的电力通信网络中。

With the development of society and the increasing demand for electricity,to guarantee the transmission reliability of important data in electric power communication networks has become one of the main focuses.In particular,when facing the changes in network topology due to natural disasters,human social activities,etc.,as well as the variations in communication requirements of electric power services,it is very challenging to quickly establish reliable routes for all services.It is difficult to obtain the optimal path combination in real time through traditional methods.In this paper,by analyzing the characteristics of network structure and task requirements,we formulate a routing optimization problem for both dual-routing and single-routing applications,aiming to improving the transmission reliability of electric power data.By combining the graph neural network with depth-first search method,we propose a novel routing scheme which is scalable for different network scenarios and service requirements.Simulation results show that compared with the traditional routing algorithm,our proposed algorithm can achieve higher routing reliability under delay and capacity constraints,and can be extended to the power communication network with different structures.