Application of Prediction Model in Monitoring LAN Data Flow Based on Grey BP Neural Network

Abstract

The grey model is characterized as less data and fast computing, and prediction results of BP neural network for nonlinear systems are better. Through amending grey prediction model by BP neural network, the grey BP neural network model is established, which improves the prediction accuracy of LAN data flow. Collecting the actual LAN data flow in a certain period, the simulation experiments of grey BP neural network model, a single grey model as well as a single BP neural network model are carried out. Experimental results show that the grey BP neural network model is superior to that of a single model. Grey BP neural network model can give full play to the advantages of each single model, but the model can not avoid the shortcomings of a single model. in the grey BP neural network model, because a grey model predicts the system trend very well only in a adjacent period, the prediction accuracy with time lasting will deviate from the prediction of actual sequence. Based on short-term LAN Data Flow, grey BP neural network model is adopted by the grey prediction characteristics. In conclusion, grey BP real-time prediction model can forecast and be able to guarantee the prediction accuracy, which determines a reasonable number of samples so that the accuracy of neural network model is further improved and, at the same time, ensure the rapid speed of operation.