信息安全研究 ›› 2020, Vol. 6 ›› Issue (6): 0-0.

• 检测预警与态势感知专题 •    下一篇

基于MEA-LVQ的网络态势预测模型

张泽1,樊江伟1,周南2   

  1. 1. 中国人民银行兰州中心支行
    2. 中国人民银行东乡县支行 兰州大学管理学院
  • 收稿日期:2020-06-08 出版日期:2020-06-05 发布日期:2020-06-09
  • 通讯作者: 张泽

Network Situation Prediction Model Based on MEA-LVQ

  • Received:2020-06-08 Online:2020-06-05 Published:2020-06-09

摘要: 网络安全事件时刻发生,互联网持续处于危险之中,网络安全时刻受到威胁.网络安全态势是评估一段时间内的网络安全状态的指标,为预防网络安全事件的发生提供了前提条件.归回型神经网络常用来解决网络安全态势评估问题,但是模型存在许多缺陷,导致预测准确率不高,为了提高分类准确率,建立MEA-LVQ的网络态势预测模型,使用思维进化算法优化网络的初始权值可以有效弥补LVQ神经网络的缺陷.采用数据集进行5次实验,模型每次分类的准确率均在90%以上,实验结果表明该模型可有效处理网络安全态势的分类问题,具有较好的评估分类能力,可以为管理人员提供较为可靠的参考值,管理人员可以及时处理威胁网络安全的事件,有效维护网络的安全与稳定.

关键词: 网络安全, 态势预测, 思维进化算法, 学习向量量化, 神经网络

Abstract: Network security incidents happen all the time, and the Internet continues to be in danger, The network security situation is an index to evaluate the network security situation in a period of time, which provides a precondition for preventing the occurrence of network security incidents. The regression neural network is often used to solve the problem of network security situation evaluation, but there are many defects in the model, resulting in low prediction accuracy. In order to improve the classification accuracy, network situation prediction model based on MEA-LVQ is established, using the mind evolutionary algorithm to optimize the initial weights of the network can effectively make up for the shortcomings of LVQ neural network. Five experiments are carried out with datasets, and the accuracy of each classification of the model is more than 90%. The experiment results show that the model can effectively deal with the classification problem of network security situation, has better evaluation and classification ability, and can provide managers with more reliable reference value, managers can deal with the incidents threatening the network security in time, and effectively maintain the network security and stability.

Key words: network security, network situation prediction, mind evolutionary algorithm, learning vector quantization, neural network