一种基于累加PSO-SVM的网络安全态势预测模型

Network security situation prediction method based on PSO-SVM

在网络安全态势感知系统中,态势预测是关键的环节。为了保证及提高态势预测的精度,结合粒子群算法的寻优性能好和支持向量机的预测准确的优势,提出了一种在数据累加预处理基础上的PSO-SVM预测模型。此模型利用将原始序列累加,弱化了原始序列中的不规则扰动影响,增强了序列的规律性的特点,与粒子群优化支持向量机(PSO-SVM)相结合,更好地发挥预测精度高的优势,更能保证预测精度。通过仿真实验检验此模型的有效性,并与PSO-SVM预测模型的结果进行对比,验证了其预测精度的优越性。

In the whole network security situational sensing system, the prediction of network security situation is the key link. In order to guarantee and improve the precision of prediction of the situation, this paper combined the particle swarm op- timization （PSO） which owned good performance in finding optimization with the supported vector machine （SVM） which had the advantage of accurate prediction. Thus, it proposed a prediction model based on accumulative PSO-SVM. It accumulated the original sequence by this model, and enhanced the weakening of the irregularity in the original sequence disturbance, and the characteristics of regularity of the sequence. In additional, the model combines with PSO-SVM better show the advantage of high prediction accuracy, ensuring the prediction precision. Simulation comparison verifies the superiority of high prediction accuracy.