近红外光谱智能分析方法研究进展OA

Near-infrared spectroscopy(NIRS)has been widely applied in many fields owing to its advantages of rapid analysis,nondestructive measurement,simple sample pretreatment,and suitability for online detection.With the increasing demand for complex sample analysis,NIRS research was expanding from traditional workflow optimization,including preprocessing,variable selection,and calibration modeling,toward mechanistic interpretation,spatial-spectral fusion,and emerging intelligent analytical methods.This paper reviewed recent advances in water spectral probes,hyperspectral imaging,large language models and spectral analysis agents,quantum computing,and quantum-inspired methods.Special attention was given to the role of water-structure responses in the mechanistic interpretation of near-infrared information and their value in revealing intermolecular interactions and information sources in complex systems.The advantages of hyperspectral imaging in the joint spatial-spectral characterization of heterogeneous samples were discussed,highlighting its importance in improving sample identification,localization,and quantitative analysis.In addition,the potential of spectral analysis agents in knowledge organization,workflow coordination,and modeling support was summarized,and the role of large language models in promoting spectral analysis from single-step modeling toward knowledge-enhanced and multi-tool collaborative analysis was examined.The potential value of quantum computing and quantum-inspired methods in spectral simulation,data augmentation,high-dimensional feature mapping,variable selection,and spatial-spectral data processing was also discussed.Finally,future development trends in mechanistic interpretation,spatial-spectral fusion,and emerging intelligent analytical methods for NIRS were discussed to provide a useful reference for related research.