Network abnormal traffic detection method based on multi kernel KPCA-PSO-ELM

In order to improve the detection rate of network abnormal traffic and reduce the false alarm rate, a network abnormal traffic detection method based on multi-core kpca-pso-elm is proposed. Firstly, Gauss radial basis function and d-order polynomial kernel function are mixed into multi-core kernel function, which has the generalization ability and learning ability of global and local kernel functions at the same time, so as to improve the efficiency of anomaly detection; Then, particle swarm optimization (PSO) is used to optimize the initial parameters of extreme learning machine (ELM), obtain the optimal solution of the parameters, reduce the influence of blindness in the process of random selection of parameters on the detection performance of limit learning machine, and improve the accuracy of anomaly detection. Finally, the optimized elm model is used to detect and classify the data. In order to verify the effectiveness of this algorithm, MATLAB and KDDCUP99 data set are used for experimental verification. The experimental results show that the detection effect of multi-core kpca-pso-elm model is significantly improved compared with PSO-SVM and PCA-BiGAN, and effectively improves the detection efficiency of network abnormal traffic detection.

I. Introduction

The rapid development of Internet technology has made network security issues increasingly prominent. Abnormal traffic detection is one of the effective means to discover potential network threats and reduce the harm of network attacks [1]. Network traffic anomaly detection technology is to extract useful network traffic identification features from the data information flow, and build and establish classification models through modeling methods. According to these classification models, the network traffic can be analyzed and evaluated as to whether the network traffic is normal or not, and finally given results of abnormal traffic identification.