蚁群算法优化RBF神经网络的网络流量预测

传统RBF神经网络在网络流量预测过程中存在收敛速度慢、极易出现局部最优等缺点,从而导致预测精度低。采用蚁群算法优化RBF神经网络参数来进行网络流量预测。利用蚁群优化算法来训练RBF神经网络的基函数宽度和中心,简化网络结构,加快收敛速度,防止局部最优的出现,改善RBF神经网络的泛化能力。实验结果表明,相对于GA-RBF以及PSO-RBF流量预测模型,模型预测准确度更高,能够很好地描述网络流的变化规律。具有泛化能力强、稳定性良好的特点,在网络流量预测中有一定的实用价值。

Network traffic prediction based on neural network optimized by ACO

Traditional RBF neural network in the forecasting process of the network traffic convergence is slow, prone to local optima and other shortcomings, resulting in low prediction accuracy and difficult problems. In order to improve the prediction accuracy of network traffic, network traffic prediction method for an ant colony algorithm to optimize the parameters of RBF neural network. Ant colony algorithm for training RBF neural network center and the width of the basis functions to speed up the convergence, simplifying the network structure to improve the generalization capability of RBF neural networks, and optimized RBF neural network to predict the network traffic, to prevent local optimum to appear. The experimental results show that relative to the commonly used BP neural network traffic prediction model, model predictions more accurate, can be well described by the variation of the network flow. With good generalization ability, good stability and a certain practical value in the prediction of network traffic.