摘要
利用BP神经网络的函数逼近特性和GM(2,1)的模型对非单调变化规律数据的描述特性,构建了一种新型的GNNM(2,1)模型,并将其用于粮仓温湿度变化趋势的建模。采用某粮仓一年内的温湿度采样值对其进行训练,并用训练好的模型预测来年温湿度的变化趋势。仿真结果表明,该模型的预测值有较高的精度,这对保证粮食储备安全具有一定的实用意义。
In the article,the function approximation characteristics of the BP neural network and the non - mon- otone description characteristics of the GM (2,1) were combined in a new model - the new GNNM (2,1) model. And it is used for modeling of the granary temperature and humidity. It was trained by the temperature and humidity sampling data of a granary in a year. The trained model was used in the prediction of the temperature and humidity of the next year. The simulation results showed that the it had the high accuracy. There are certain practical significance for food reserves security.
出处
《中国粮油学报》
EI
CAS
CSCD
北大核心
2012年第12期108-110,共3页
Journal of the Chinese Cereals and Oils Association