The Prediction Of Flow Rate Of Campus Network Bandwidth Based On BP Neural Network

At present, the behavioral characteristics of the network are getting complicated day by day, and it's a gigantic challenge to program and administrate the networks as the development of Internet. Now it's necessary to analyze the flow of the network and build the effective model to carry out the flow of the network. The flow model of the network is to depict the properties of the network, not only reflecting the history rate of flow properties accurately, but also forecasting the rate of flow in the network for some time. In this paper we use the non-linear model to improve the forecast accuracy and adaptive capacity, and on this basis, we design and realize the network traffic analysis system.This paper theoretically studies the conception and algorithm of the BP neural network model , and researches the technology of the forecast information and the feasibility of traffic prediction using the artificial neural network time series based on the campus network management practice. On this basis, a bandwidth prediction model based on BP neural network is explored.The BP neural network's prediction backup problems in time series are discussed on the principle of changing of the characteristics of bandwidth. And the neural network prediction topological model and other network's parameters are designed by using formula and the method of experiment, and the adaptive learning rate adjustment and adding momentum to the two BP improved algorithm are given.The network traffic forecasting system is designed and realized by combining the BP algorithm and time series. The results are forecasted and analyzed. We obtain the optimal bandwidth share forecast model structure by using different BP algorithm to train and test the neural network . And then the short-term changes in network traffic application model can be predicted, obtaining higher accuracy.