论文部分内容阅读
对混胺燃料近红外光谱进行光谱预处理和波长选择,采用偏最小二乘法建立各分析指标的定量校正模型,利用校正集标准偏差(SEC)和预测集标准偏差(SEP)比较光谱预处理和波段选择对各项指标分析模型的影响。平滑处理对各项指标的影响不大,一般采用5点平滑。由于混胺颜色的影响,二阶微分比一阶微分预处理光谱的效果要好。波段选择对各分析指标的影响差别较大,应综合考虑各组成含氢基团的贡献和其它成分共线性的影响,选择合适的波长范围。提出了建立高精度、高稳健性分析模型应注意的问题。
Spectral preprocessing and wavelength selection of the near infrared spectra of the mixed amine fuel were carried out. The quantitative correction model of each analytical index was established by the partial least squares method. The spectral pretreatment and the standard deviation of the prediction set (SEP) Influence of band selection on various index analysis models. Smoothing has little effect on the indicators, generally 5 smooth. Due to the influence of the mixed amine color, the second-order differential is better than the first-order differential preconditioned spectrum. The influence of band selection on each analytical index is quite different. The influence of the contribution of each group of hydrogen-containing groups and the collinearity of other components should be comprehensively considered, and the appropriate wavelength range should be selected. The problems that should be paid attention to in establishing high-precision and high-robust analysis model are proposed.