微型电脑应用

电商网络的恶意流量中存在长序列、多任务等特征，导致检测精度下降，对此，研究基于门控循环单元(GRU)神经网络的电商网络恶意流量检测方法。在相空间重构过程中，引入小波变换消除电商网络流量中存在的噪声。采用主成分分析法对去噪后的电商网络流量进行特征提取。将经过特征提取的电商网络流量输入GRU神经网络，通过训练和优化神经网络得到电商网络恶意流量检测结果。实验结果表明，所提方法的电商网络流量处理效果好，恶意流量类别检测精度高达到99%,最短检测时间仅为234 s,具有一定的技术水平与实用性。

There are many long sequence and multitasking features in malicious traffic in e-commerce networks, leading to a decrease in detection accuracy. Therefore, a malicious traffic detection method based on gated recurrent unit(GRU) neural network in e-commerce networks is studied. In the process of phase space reconstruction, wavelet transform is introduced to eliminate the noise in e-commerce network traffic. The principal component analysis method is used to extract the features of the e-commerce network traffic after denoising. The e-commerce network traffic after feature extraction is input into GRU neural network, and the malicious traffic detection results of e-commerce network are obtained by training and optimizing the neural network. Experimental results show that the e-commerce network traffic processing effect of the proposed method is good, the detection accuracy of malicious traffic category is as high as 99%, and the minimum detection time is only 234 s, which has a certain technical level and practicality.