摘要
将人工神经网络（ａｒｔｉｆｉｃｉａｌｎｅｕｒａｌｎｅｔｗｏｒｋｓ）应用于喇曼光谱数据处理中。研究了学习时间常数μ（ｌｅａｒｎｉｎｇｒａｔｅｃｏｅｆｆｉｃｉｅｎｔｓ）及传递函数（ｔｒａｎｓｆｅｒｆｕｎｃｔｉｏｎｓ）对网络性能的影响，发现当μ＝０．５时，网络运行最佳。通过比较原始谱图及经网络处理后所得谱图，证明采用带有Ｓ形传递函数的前向网络能获得较好的信噪比，谱线的分辨率也有所提高。
Artificial neural networks and its application in Raman spectra data processing are described. Studying the back-propagation neural networks with different rate coefficients μand transfer functions, we found that the networks. peformed optimally at μ= 0.5. Comparing the original and processed spectra,we have shown that the networks with sigmoid transfer functions can correct minor spectral distortions,give better signal-to-noise ratios,and somewhat improve the spectral resolution.
出处
《分析测试学报》
CAS
CSCD
1996年第3期1-6,共6页
Journal of Instrumental Analysis
