首页|期刊导航|徐州工程学院学报(自然科学版)|改进RF-POA数学模型在不平衡大数据网络攻击诊断中的应用

改进RF-POA数学模型在不平衡大数据网络攻击诊断中的应用OA

Application of an Improved RF-POA Mathematical Model in Network Attack Diagnosis for Imbalanced Big Data

中文摘要英文摘要

针对不平衡大数据场景下网络攻击诊断任务中误报率与漏报率居高不下等问题,提出一种基于改进随机森林鹈鹕优化算法(random forest-pelican optimization algorithm,RF-POA)的不平衡大数据网络攻击诊断数学模型.通过引入 POA 对 RF 模型的关键参数进行自动化寻优,以提升模型的分类性能与泛化能力.利用 SMOTE 方法对原始不均衡网络流量数据进行平衡处理,克服类别分布不均对诊断的影响.进而构建融合时序特征的攻击诊断框架,执行特征加权与决策树二分策略实现精准分类.基于 CICIDS2017 数据集进行实验验证,结果表明:RF-POA 模型误报率稳定在2.0%~3.0%,且消融实验进一步证实了算法改进的有效性.由此得出结论,改进模型为不平衡大数据场景下的网络攻击诊断提供了一种有效的解决方案.

To address the persistent issues of high false alarm and missed detection rates in diagnosing network attacks under imbalanced big data scenarios,this paper proposes an improved mathematical model based on the Random Forest-Pelican Optimization Algorithm(RF-POA).The POA algorithm automatically optimizes the key parameters of the random forest model,thereby enhancing its classification performance and generalization capability.The synthetic minority over-sampling technique(SMOT E)balances the original imbalanced network traffic data,mitigating the adverse effects of uneven class distribution on diagnostic accuracy.Furthermore,an attack diagnosis framework that integrates temporal features has been constructed.Within this framework,feature weighting and a decision tree binary splitting strategy are applied to achieve precise classification.Experiments conducted on the CICIDS2017 dataset demonstrate that the RF-POA model maintains a stable false alarm rate within the range of 2.0~3.0%.Ablation experiments further confirm the effectiveness of the proposed algorithmic improvements.Therefore,it can be concluded that the improved model effectively addresses the challenge of diagnosing network attacks in imbalanced big data environments.

张雪婷

铜陵职业技术学院,安徽 铜陵 244000

信息技术与安全科学

攻击诊断不平衡大数据随机森林算法鹈鹕算法

attack diagnosisimbalanced big datarandom forest algorithmpelican optimization algorithm

《徐州工程学院学报(自然科学版)》 2026 (3)

50-58,9

铜陵职业技术学院校级科研自然重点项目(TZY25ZRZD06)安徽省高等学校质量工程教学研究一般项目(2025jyxm0871)

评论