摘要
研究了用于红外光谱碎片结构识别的新型ＢＰ网络模型，并由此构建了ＰＲＯＧＡＮＮ程序软件，提出了红外光谱数据的预处理以及网络输出的统计分析方法。通过对４４种结构碎片的识别结果表明，所提出的模型与方法较之常规ＢＰ网络具有更大优越性。
A novel BP neural network model was systemlly investigated as a method for structure identification of infrared spectra and the software PROGANN was constructed with this model. The data-preprocessing method and the statistical analysis of output results for NN was presented in detail. The predicted for 44 structural fragments showed that this model had a better results than conventional BP neural networks.
出处
《抚顺石油学院学报》
CAS
1996年第3期37-42,共6页
Journal of Fushun Petroleum Institute
基金
国家自然科学基金
国家教委优秀年轻教师基金
关键词
神经网络
BP算法
红外光谱
结构识别
Neural networks
BP algorithm
Infrared spectrometry
Structure identification
