分段直接校正(PDS)算法是目前最常用的近红外光谱模型传递方法,但它在对整个谱区进行校正时,始终依赖大小不变的传递窗口.为了提高传递效果,本研究在PDS基础上提出了一种新的算法--小波多尺度分段直接校正法(WMPDS),用于混胺的近红外光谱模型传递,并详细讨论了模型的传递参数和传递结果.本算法首先对混胺的近红外光谱进行小波分解,然后用PDS算法对每一层小波系数进行传递,PDS窗口随小波系数频率的提高进行动态调整,最后进行小波重构.本算法能有效消除不同仪器之间的大部分差异,大幅度地改善分析精度,与传统的PDS相比,传递效果明显提高.%Piecewise direct standardization (PDS) is the most popular method on near infrared analysis calibration transfer. One of the disadvantages of PDS is that the window widths are the same in the whole near infrared (NIR) spectral wavelengths as the window width is too large in some spectra region and too small in the others. A new calibration transfer method based on PDS was proposed to improve the calibrated results, the new method is wavelet multi-scale piecewise direct standardization (WMPDS). It was used for calibration transfer in NIR analysis of mixed amine. Wavelet transfer (WT) was conducted to discompose NIR data and then PDS was applied to correct the wavelet coefficients. The window widths of PDS were changed as the wavelet coefficients changed and the reconstruction of the coefficient was next to the correction of PDS. WMPDS can well correct the differences between instruments and improve analytical accuracy greatly. The performance of WMPDS is better than that of PDS in NIR calibration transfer.
展开▼
