OLAM网络分析水泥生料X荧光谱中学习谱的选择  

Learning Spectrum's Selection in OLAM Network for Analysis Cement Samples

作　　者：黄宁[1] 王鹏[1] 唐代全[1] 胡纫兰[1] 

机构地区：[1]四川大学原子核科学技术研究所,成都610064

出　　处：《核电子学与探测技术》2010年第1期93-95,70,共4页Nuclear Electronics & Detection Technology

摘　　要：利用最优线性联想记忆(OLAM)神经网络方法对水泥生料样品X荧光谱进行了分析,学习谱的选择采用了两种方式:单质谱学习和混合谱学习。对这两种方式及输出结果作了对比,单质谱学习可以构建模拟谱,用于对谱漂移和漏学习的判别;混合谱学习具有信息容量大,考虑了基体效应的优点,但是学习谱之间的多重相关性可能对结果造成一定影响。It uses OLAM artificial neural network to analyze the samples of cement raw material Two kinds of spectrums are used for network learning, pure-element spectrum and mix-element spectrum. The output of pure-element method can be used to construct a simulate spectrum, which can be compared with the original spectrum and judge the shift of spectrum; the mix-element method can store more message and correct the matrix effect, but the multicollinearity among spectrums can cause some side effect to the results.

关 键 词：OLAM XRF 水泥生料样品 学习谱选择 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]