基于优化ELM的信息物理融合系统入侵检测  

Cyber physical system intrusion detection method based on optimized ELM

作　　者：肖韵婷 张立臣[1] XIAO Yun-ting;ZHANG Li-chen(School of Computer Science,Guangdong University of Technology,Guangzhou 510006,China)

机构地区：[1]广东工业大学计算机学院,广东广州510006

出　　处：《计算机工程与设计》2022年第1期9-17,共9页Computer Engineering and Design

基　　金：国家自然科学基金项目(61873068)。

摘　　要：针对带权正则化极限学习机(WRELM)性能受随机初始值、数据不平衡及离群点影响大的问题,提出基于局部距离的带权正则极限学习机(LDWRELM),提高对不平衡数据集与离群点的抗干扰能力,使用改进的头脑风暴优化算法(MBSO)对LDWRELM的初始权重阈值进行联合优化。MBSO在头脑风暴优化算法(BSO)的基础上对个体更新与变异策略进行改进,在典型函数上验证了该改进对全局寻优能力与收敛速度的提升。构建基于MBSO优化的LDWRELM信息物理融合系统(CPS)入侵检测模型,将仿真结果与其它算法进行比较,验证了MBSO-LDWRELM算法有效提高了准确率,降低了误报率与检测时间。Aiming at the problem that the performance of weighted regularized extreme learning machine(WRELM)is greatly affected by random given input weights and thresholds,imbalanced dataset and outliers,the local distance-based weighted regularized extreme learning machine(LDWRELM)was proposed to improve anti-interference ability of imbalanced datasets and the outliers.The modified brainstorming optimization algorithm(MBSO)was proposed to jointly optimize the initial weights and thresholds of LDWRELM.Based on the brainstorm optimization algorithm(BSO),the individual update and mutation strategy was improved and the convergence speed and global optimization capability of MBSO were verified by the test of typical functions.A cyber physical system(CPS)intrusion detection model based on LDWRELM optimized by MBSO was constructed,which was compared to other well-known algorithms.Results show that the proposed algorithm is better than several other algorithms in accuracy rate enhancement,and false alarm rate and detection time reduction.

关 键 词：信息物理融合系统 头脑风暴优化算法 带权正则化极限学习机 入侵检测 全局寻优 

分 类 号：TP393.08[自动化与计算机技术—计算机应用技术]