Particle Swarm Optimization RBF Neural Network Model for Internet Traffic Prediction

A RBF-based neural network adaptive particle swarm optimization algorithm is proposed in this paper. In this algorithm, code at particle position adopts quantum bit to realize. The paper adopt particle flight path information to dynamically update the status of quantum bit and introduces quantum non-gate to realize mutation operation so as to avoid local optimization. Then, it is used to train neural network, realize radial basis function to optimize neural network parameter and establish self-adaptive PSO-RBF neural network algorithm-based network traffic prediction model. The prediction results of real network traffic show that the convergence speed and prediction accuracy in this method are superior to traditional RBF neural network, particle swarm RBF neural network, hybrid particle swarm RBF neural network and adaptive particle swarm RBF neural network. Meanwhile, predicting effect is hardly affected by time scale change.