摘要
灰色建模要求的样本点少,不必有较好的分布规律,而且计算量少,操作简便。而BP网络学习样本时,会反馈校正输出的误差,具有并行计算、分布式信息存储、强容错力、自适应学习功能等优点。本文将灰色预测建模和神经网络技术融合起来,建立灰色神经网络模型(GNNM)。提出计算残差序列和新的预测值的公式。用于发酵动力学预测,结果表明,灰色神经网络模型在预测精度方面优于常规灰色模型。该模型的算法概念明确,计算简便,有较高的拟合和预测精度,拓宽了灰色模型的应用范围。
The foundation of grey model GM （1,1） need fewer samples that without better distributing order, but also do less calculating quantity. BP network adopts the back propagation when it learns samples, so it has ability of parallel calculation, distributed information memory, admitting-error, and so on. In this paper, we combine grey prediction method with the neural networks method to found the grey neural network model shortly named GNNM. The equations are presented for calculating residual sequence of the GM （ 1, 1 ） and new predicted values. GNNM is used for fermentation kinetics. The suitable conditions of back-propagation （BP） neural network are：input nodes：3 （it is equal to predicted order） ; hidden nodes ：6; output node：1; the learning epochs：200. Predicted results indicated that the grey neural network model in quality were better than the normal grey model GM （ 1,1 ） . The practical example show that GNNM is definite in concept, convenient in calculation, good in fitting and precise in prediction, thus GNNM improves the precision of the GM （ 1,1 ） model and enlarges its application scope.
出处
《计算机与应用化学》
CAS
CSCD
北大核心
2007年第8期1078-1080,共3页
Computers and Applied Chemistry
关键词
灰色理论
神经网络
灰色神经网络模型
发酵动力学
grey theory, neural network, grey neural network model（GNNM） , fermentation kinetics
