Network Traffic Prediction Based on Improved GA-Elman Neural Network

Abstract

Aiming at the current demand of network traffic modeling and prediction, an Elman network prediction model optimized by improved GA algorithm is proposed. Firstly, the structure of the standard Elman neural network is improved by adding input undertake layer and output undertake layer. Considering that the crossover and mutation operations in the genetic algorithm can easily affect the algorithm performance, the adaptive crossover and mutation formulas are proposed, and the improved GA-Elman neural network is derived, and the optimal network parameters are obtained. Finally, the network traffic time series data are used to forecast. The experimental results of three Elman network prediction models show that the method presented in this paper can predict network traffic with high accuracy and has certain practical value in network traffic prediction.