Near-infrared model transfer of wheat flour based on reverse standardization combined with stable and consistent...

The aim is to transfer near-infrared models for wheat flour protein content between two portable near-infrared spectrometers: the IAS-5000 and the VIAVI, this study proposes a calibration method combining reverse standardization (RS) with a screening wavelengths with consistent and stable signals (SWCSS). First, RS is applied to the master spectra using the slave spectra as reference. Then, the SWCSS algorithm filters consistent wavelengths between the transformed master and slave spectra, followed by hybrid modeling from both instruments. A quantitative model for the slave instrument is built within the framework of partial least squares regression (PLSR) and compared with spectral subtraction correction (SSC), direct standardization (DS), and piecewise direct standardization (PDS). Results show that for instruments with significant spectral differences, the direct reverse standardization (DRS) model achieves superior transfer performance over reverse subtraction correction (RSC) and piecewise reverse standardization (PRS). The combination of reverse standardization, SWCSS, and hybrid modeling fully exploits the flexibility of RS for model reconstruction, achieving superior predictive performance with an RPD value of 3.94. The results indicate that model transfer performance is jointly influenced by spectral alignment, wavelength stability, and calibration sample distribution. By integrating these factors within a unified framework, the proposed method provides a more robust and flexible solution for near-infrared calibration transfer, offering improved reliability for practical multi-instrument applications.