BP Artificial Neural Network Based on Predictive Model for Pulmonary Heart Disease
Objective: To study the application of BP artificial neural network tools for the forecasting of pulmonary heart disease incidence rate which exists auto-regressive and moving average phenomenon. Methods: First, time series analysis was adopted, the network input variables included AR (1), MA (1), MA (2), MA (3), MA (4), average air temperature, average air pressure and relative humidity. The output of network is a transform value of the incidence rate of viral hepatitis neural network tools box v4.0.2 of Software MATLAB 7.0 was used to structure, train and simulate BP Artificial Neural Network. The data from 2003 to 2009 were used as a training set and the data from 2009 made up the test set. Results: The application of the BP artificial neural network enabled the RNL of 0.918958, while the RNL of linear model is 0.673104. Conclusion: BP artificial neural network is superior to conventional methods in solving problem which exist Auto-regressive and moving average phenomenon.
back propagation artificial neural network leverberg-maquard auto-regressive moving average time series analysis pulmonary heart disease
TIAN Fupeng MA Liangliang
School of Computer and Information, Northwest University for Nationalities, P.R.China, 730030
国际会议
威海
英文
430-433
2010-07-24（万方平台首次上网日期，不代表论文的发表时间）