摘要
介绍对动物药阿胶进行了化学分析,利用BP型人工神经网络对各类阿胶进行模式识别,并进行神经网络的容错分析,结果说明,BP网络有相当强的预测能力和容错能力,当隐含层损坏超过20％时,预测能力才会明显下降。
Chemical analysis and pattern recognition of various donkeyhide glue through BP artificial neural network is performed, and compatible analysis of neural network through decreasing the number of the hide layers is followed. The result proves that BP network has strong forecast capacity and compatible capacity. Only when the damage of hide layers is beyond 20% , the forecast rate has a evident drop.
出处
《计算机与应用化学》
CAS
CSCD
北大核心
2002年第4期411-414,共4页
Computers and Applied Chemistry
基金
河南省杰出青年基金(99-13)
