小波变换-分段直接校正法用于近红外光谱模型传递研究

提出了一种新的传递算法(WT-PDS)———小波变换-分段直接校正法,并详细讨论了模型传递参数和传递结果。首先利用小波变换对光谱进行压缩处理,采用PDS算法消除不同仪器之间压缩数据的差异,最后利用经校正的压缩数据进行分析,实现模型传递。本方法能够扣除不同仪器之间的大部分差异,大幅度改善分析精度。传递后模型分析精度与源机模型稳健性紧密相关。如果源机模型稳健性强,则能够实现不同仪器之间的共享。本方法能够实现源机的0#轻柴十六烷值、凝点、馏出温度;-10#轻柴十六烷值、凝点以及-10#军柴凝点和馏出温度共10个模型在5台仪器之间共享,简化了建模的成本。与传统的PDS相比,WT-PDS方法具有传递和建模变量少、速度快、光谱校正性能高等优点,而其模型分析精度与传统PDS基本一致。

Application of Wavelet Transform-Piecewise Direct Standardization on the Near Infrared Analysis Model Transfer

This study proposes a new algorithm,wavelet transform-piecewise direct standardization(WT-PDS),used for model transfer in near infrared(NIR) analysis.Wavelet transform(WT) is conducted to compress NIR(data) and then piecewise direct standardization(PDS) algorithm is applied to correct compressed NIR data differences between instruments.WT-PDS is based on linear algorithm and can well correct the linear differences between instruments and improve analytical accuracy greatly.The precise of slave model transferred by WT-PDS depends in part on the robustness of master model.Several models of master instruments,such as cetane numbers,solidification point,(distillation) temperature of 0~(#) light diesel,cetane number,solidification point of -10~(#) light diesel and solidification point,distillation temperature of -10~(#) military diesel can be transferred and used in four slave instruments.The small variables of WT-PDS can reduce time of model transferring and building,without deterioration of analytical accuracy.The performance of WT-PDS is better than that of traditional method,classical pitcewise direct standardization(CPDS) in NIR spectra correction.