Establishment on Early Warning System of Grain Security in Zhangjiagang, China Using Immune-Based Optimized BP Neural Network Model

Abstract

To solve the problems of poor accuracy and greater fluctuations in the grain output forecast, this paper introduces artificial immune algorithm used in BP neural network for training neural network. Results show improved BP neural network (IBOA) overcomes the conventional BP neural network in the aspects of slow convergence and inefficiency with better performance for the forecast and accuracy. The model is used to forecast grain output in Zhangjiagang city, representative of medium-sized cities in developing countries. Forecasting results show the total output of wheat and rice would has an increasing trend from 2010 to 2013 except some unpredictable factors. IBOA model can be used as a better method of grain security early warning instead of the conventional BP model to provide policy guidance for local government.