基于改进人工鱼群的工业互联网异常检测研究

The anomaly detection analysis on the industrial Internet based on improved artificial fish swarm

针对ICS工控网络异常检测存在的问题,在传统ELM网络异常检测模型的基础上,引入改进人工鱼群算法的ELM工控网络异常检测模型。首先对样本进行归一化处理和K-means均值聚类;其次基于惯性计算思想对人工鱼群算法中的聚集行为和觅食行为进行改进,通过比较惯性计算后鱼群中心位置与鱼个体当前位置的适应度值来判定人工鱼群的活动情况,输出最优权值和阈值作为ELM的输入;最后进行异常检测训练。实验测试结果表明,改进算法易跳出局部最优解,对攻击类数据的检出率达到90%以上。

Aiming at the problem of ICs industrial control network anomaly detection,the ELM industrial control network anomaly detection model based on improved artificial fish swarm algorithm is introduced.Firstly,the samples are normalized and K-means clustering is carried out;then the idea of inertia calculation is introduced to improve the aggregation behavior and foraging behavior in the artificial fish swarm algorithm.The execution behavior of artificial fish is judged by comparing the fitness value of the center position and the current position after inertia calculation,and the optimal weight and threshold are output as the input of elm for anomaly detection training.The experimental results show that the algorithm constructed in this paper is easy to jump out of the problem of local optimal solution,and the detection rate of the improved algorithm for attack types is the highest,which reaches more than 90%,and realizes the detection of abnormal data in industrial control network.