基于深度强化学习的电力巡检机器人网络自动化监测系统  

Network Automation Monitoring System for Electric Power Inspection Robots Based on Deep Reinforcement Learning

作　　者：陈人楷 方晓明 李仕彦 CHEN Renkai;FANG Xiaoming;LI Shiyan(Information and Communication Branch,State Grid Fujian Electric Power Co.,Ltd.,Fuzhou 350001,China)

机构地区：[1]国网福建省电力有限公司信息通信分公司,福州350001

出　　处：《自动化与仪表》2024年第9期70-73,83,共5页Automation & Instrumentation

摘　　要：为提高网络巡检数据传输的可靠性与时效性,设计基于深度强化学习的电力巡检机器人网络自动化监测系统。通过系统的数据采集模块,获得目标机器人网络的历史数据与实时数据,共同存入原始采集数据库;由数据处理模块处理该数据库内数据,降低冗余,提升数据精度;结合自动化监测模块的漏洞检测扫描器,由处理后历史数据中获得异常数据,存入历史异常数据库,调用该模块的深度强化学习算法,以历史异常数据为输入,实现对目标网络实时数据的自动化监测。实验结果显示,系统的数据传输误码率低、传输时延较短、数据传输的综合性能较好;可实现目标网络历史数据与实时数据的异常监测,监测结果的准确度高于97%,且监测用时短、时效性较佳、综合性能表现优越。The automatic monitoring system of electric power inspection robot communication network based on deep reinforcement learning is studied to improve the reliability and timeliness of communication network inspection data transmission.Through the data acquisition module of the system,the historical data and real-time data of the target robot communication network are obtained and stored in the original acquisition database together.The data in the database is processed by the data processing module to reduce redundancy and improve data accuracy.Combined with the vulnerability detection scanner of the automatic monitoring module,the abnormal data is obtained from the processed historical data,stored in the historical anomaly database,and the deep reinforcement learning algorithm of the module is invoked to realize the automatic monitoring of the real-time data of the target communication network with the historical anomaly data as input.The results show that the system has low bit error rate,short transmission delay and good comprehensive performance of data transmission.It can realize anomaly monitoring of historical data and real-time data of target communication network,and the accuracy of monitoring results is higher than 97%,and the monitoring time is short,the timeliness is better,and the comprehensive performance is superior.

关 键 词：深度强化学习 电力巡检机器人 网络自动化监测 漏洞检测 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置] TP242[自动化与计算机技术—控制科学与工程]