[Influence of FT-NIR spectrometer scanning requirements on the math model's precision]

Area of Science:

• Agricultural Science
• Analytical Chemistry
• Spectroscopy

Context:

• Near-infrared (NIR) spectra databases are crucial for agriculture product quality analysis and breeding.
• Maximizing the utility of these foundational databases requires optimizing spectral data acquisition.
• Standardizing scanning parameters is essential for sharing NIR resources effectively.

Purpose:

• To investigate the influence of varying spectral resolution, He-Ne frequency, and sample granularity on quantitative models for wheat powder protein.
• To determine the optimal parameters for obtaining high-quality NIR spectra for agricultural applications.

Summary:

• This study analyzed the impact of different resolutions (4, 8, 16 cm⁻¹), He-Ne frequencies, and sample granularities on wheat powder protein prediction models using NIR spectroscopy.
• Results indicate that spectral resolution has a negligible effect on the protein model.
• While He-Ne frequency shifts affect wavenumber accuracy, the impact is insignificant within a 1 cm⁻¹ range, suggesting infrequent adjustments for FT-NIR instrument stability.
• Sample granularity emerged as a critical factor significantly influencing the accuracy of NIR-based chemometric models.

Impact:

• Provides practical guidance for researchers and users of NIR spectroscopy in agriculture to enhance data quality and model reliability.
• Highlights the importance of controlling sample granularity for accurate agricultural product analysis.
• Contributes to the standardization of NIR data acquisition for improved resource sharing and application in agriculture.