改进的神经网络在网络流量预测中的应用研究

摘要

Network traffic prediction is studied. Network flow data has the characteristics of bursty, nonlinear and time-varying, and the traditional RBF neural network has slow convergence speed and is easy to get into local minimum in the network traffic prediction process, which causes low forecasting accuracy and affects the network traffic management efficiency. In order to improve the prediction precision of network traffic, a network traffic prediction method is put forward based on particle swarm optimization algorithm and RBF neural network. Firstly, the method used particle swarm optimization algorithm to optimized the RBF neural network parameters and simplify network structure and improved the convergence speed. Then, the optimized RBF neural network was used to predict the network traffic and prevent local optimum. Finally, the proposed method was tested based on the network traffic data in Matlab platform. The experimental results show that this algorithm can improve the network traffic prediction accuracy. Compared with the traditional network traffic prediction method, this algorithm is more suitable for complicated network data flow prediction.%关于保证网络安全服务,研究网络流量预测问题.网络流量数据具突发性、非线性和时变性等等特点,传统RBF神经网络在网络流量预测过程存在敛速度慢、极易出现局部最优等缺点,从而导致预测精度低和难问题.为了提高网络流量的预测精度,提出一种粒子群算法优化RBF神经网络参数的网络流量预测方法.首先采用粒子群算法对RBF神经网络的参数进行优化,简化网络结构,加快收敛速度,并用优化后RBF神经网络对网络流量进行预测,防止局部最优的出现.最后在Matlab平台对模型进行了仿真,结果表明,算法提高了网络流量的预测精度.相对于传统的网络流量预测方法,提高了预测效率.