Research on Combined Prediction Model Based on Bp Neural Network and Its Application

In order to study the prediction ability of combined model based on BP neural network, the information is gathered to established BP neural network model. The BP neural network model is built between the incidence seniority and the influencing factors and whose prediction performance is compared with the traditional model of multivariate and linear regression. Based on this we can set up linear combination forecasting model and nonlinear combination forecasting model to find the optimal prediction model through restricting the BP neural network and multivariate linear regression method. BP neural network ,multiple linear regression ,based on the forecasting error square and minimum linear combination model ,based on the prediction error absolute value of the minimum of the linear combination model ,based on BP neural network combined model prediction of the average relative error incidence seniority is respectively 8.977%, 11.092%, 8.952%, 8.952%, 8.963%, 8.723%, the average rank respectively 2.45, 2.79,2.39,2.40,2.37. According to the error index and prediction accuracy, predicting performance can be arranged from superior to inferior in this order: BP neural network combination model, the combination forecasting to achieve the minimum square error, absolute error and minimum combination forecast, BP neural network model, multiple linear regression model, combined model prediction accuracy is better than a single model, and BP neural network combined model is the optimal.