基于弱监督学习的电力信息动态漏洞挖掘系统  

Power information dynamic vulnerability mining system based on weakly supervised learning

作　　者：王勇 裘建开 严钰君 沈涛 WANG Yong;QIU Jiankai;YAN Yujun;SHEN Tao(State Grid Ningbo Power Supply Company,Ningbo 315000,China;Ningbo Power Transmission and Transformation Construction Co.,Ltd.,Ningbo 315000,China)

机构地区：[1]国网宁波供电公司,浙江宁波315000 [2]宁波送变电建设有限公司,浙江宁波315000

出　　处：《电子设计工程》2023年第13期114-117,122,共5页Electronic Design Engineering

摘　　要：为控制电力信息漏洞数据的单位传输速率,避免漏洞信息错误挖掘行为的出现,设计基于弱监督学习的电力信息动态漏洞挖掘系统。利用Fuzzing框架,调节调试器模块与漏洞数据解析模块间的实时连接关系,完成电力信息动态漏洞挖掘系统的硬件执行环境搭建。根据弱监督学习原则,定义待处理漏洞信息所属的目标类型,联合反编译器闭环,将完成转码处理的挖掘解析语言反馈回硬件应用主机中,实现基于弱监督学习的电力信息动态漏洞挖掘系统设计。实验结果表明,在弱监督学习原则作用下,电力信息漏洞数据单位传输速率的最大值被控制在2.34 bit/s以下,与改进动态故障树的挖掘系统相比,能够较好解决现存的漏洞信息错误挖掘问题,符合实际应用需求。In order to control the unit transmission rate of power information vulnerability data and avoid the occurrence of wrong mining behavior of vulnerability information,a power information dynamic vulnerability mining system based on weakly supervised learning is designed.Using the Fuzzing framework,the real-time connection relationship between the debugger module and the vulnerability data analysis module is adjusted,and the hardware execution environment of the power information dynamic vulnerability mining system is completed.According to the principle of weakly supervised learning,define the target type of the vulnerability information to be processed,and then combine with the decompiler to close the loop,and feed back the mining analysis language after transcoding processing back to the hardware application host to realize the dynamic vulnerability mining of power information based on weakly supervised learning system design.The experimental results show that under the weak supervision learning principle,the maximum transmission rate of power information vulnerability data unit is controlled below 2.34 bit/s.Compared with the improved dynamic fault tree mining system,it can better solve the existing vulnerability information.Error mining problem,in line with practical application needs.

关 键 词：弱监督学习 电力信息 动态漏洞挖掘 Fuzzing框架 调试器 反编译器 

分 类 号：TN702[电子电信—电路与系统]