六盘水师范学院学报2025，Vol.37Issue(3)：89-97,9.DOI:10.16595/j.1671-055X.2025.03.009
无监督环境下改进Q-learning算法在网络异常诊断中的应用
lication of Improved Q-learning Algorithm in Network Anomaly Diagnosis in Unsupervised Environment
摘要
Abstract
Aiming at the shortcomings of traditional network anomaly diagnosis algorithms in unsupervised environment,such as anomaly location and low accuracy of anomaly data classification,a wireless network anomaly diagnosis method based on improved Q-learning algorithm is designed.Firstly,Data streams of wireless networks are collected based on Asynchronous Data Unit(ADU)and data packet characteristics are extracted.Then,a Q-learning algorithm model is constructed to explore the bal-ance point between the state value and the reward value,and the Simulated Annealing algorithm(SA)is used to accurately identi-fy the state at the next moment from the global view.Finally,the joint distribution probability of the training sample is deter-mined,and the approximation performance of the output value is improved to achieve the balance between the exploration and the cost.The test results show that the average accuracy of the proposed algorithm is 99.4%,which is superior to the three traditional methods in the classification accuracy and efficiency of different types of network anomalies.关键词
无监督/改进Q-learning/ADU单元/状态值/联合分布概率Key words
No supervision/Improve Q-learning/ADU unit/Status value/Joint distribution probability分类
计算机与自动化引用本文复制引用
梁西陈..无监督环境下改进Q-learning算法在网络异常诊断中的应用[J].六盘水师范学院学报,2025,37(3):89-97,9.基金项目
安徽省高等学校自然科学重点研究项目"发展新质生产力背景下云计算虚拟机安全性改进策略研究"(2024AH051836) （2024AH051836）
安徽省高校理工科教师赴企业挂职实践项目"宿州数字奇安技术有限公司"(2024jsqygz182) （2024jsqygz182）
安徽省高等学校省级质量工程项目"计算机网络基础"(2023hxkc129). （2023hxkc129）
