Near-Infrared Spectral Characteristic Extraction and Qualitative Analysis Method for Complex Multi-Component Mixtures...

Area of Science:

• Analytical Chemistry
• Spectroscopy
• Chemometrics

Background:

• Quality identification of multi-component mixtures is critical for industrial process control.
• Conventional artificial sensory evaluation is subjective, leading to inaccurate and unstable quality assessments.
• Need for objective, accurate, and stable methods for quality control in complex mixtures.

Purpose of the Study:

• To develop a novel near-infrared (NIR) spectral characteristic extraction method for quality identification.
• To establish a high-accuracy qualitative identification model for multi-component mixtures.
• To enable non-destructive, rapid quality detection and automatic process control.

Main Methods:

• Pre-processing of NIR spectra using the Norris derivative filtering algorithm for smoothing.
• Characteristic extraction via third-order tensor robust principal component analysis (TRPCA) for dimensionality reduction.
• Construction of a qualitative identification model using support vector machines (SVM).

Main Results:

• Effective reduction of raw NIR spectral data dimensionality using TRPCA.
• Achieved a high classification accuracy of 98.94% for qualitative identification.
• Demonstrated the potential for mining subtle differences between classes using low-dimensional wavebands.

Conclusions:

• The developed NIR spectral analysis method provides a non-destructive and rapid approach for quality detection.
• This method offers a robust alternative to subjective sensory evaluations for multi-component mixtures.
• The technique can be integrated into automatic quality control systems for industrial production.