研究目的
To evaluate the feasibility of applying near-infrared (NIR) spectroscopy and multivariate calibration for identifying and quantifying several common adulterants in notoginseng powder.
研究成果
The study confirms the feasibility of combining NIR spectroscopy, CARS, and PLS for quantifying common adulterants in notoginseng powder. It highlights the importance of using representative sample sets for modeling and suggests future work to focus on detecting other adulterants in notoginseng powder.
研究不足
The study acknowledges the challenges of NIR spectroscopy such as overlapping peaks and the dependence on chemometrics for constructing predictive models. The performance of models heavily depends on the partition of the dataset, and slight changes in the training set can significantly affect model performance.
1：Experimental Design and Method Selection:
The study employed NIR spectroscopy and multivariate calibration techniques, specifically partial least squares (PLS) and competitive adaptive reweighted sampling (CARS), for variable selection and model construction.
2：Sample Selection and Data Sources:
Two datasets were prepared, one with notoginseng powder adulterated with sophora flavescens powder (SFP) and corn flour (CF), and another with notoginseng powder adulterated with other analogues of low-grade (ALG).
3：List of Experimental Equipment and Materials:
An Antaris II FT-NIR spectrophotometer equipped with an integrating sphere and a standard sample accessory was used for spectral measurement.
4：Experimental Procedures and Operational Workflow:
NIR spectra were collected within the range of 4000–10000 cm?1, at
5：856 cm-1 intervals. Each final spectrum was an average of 32 scans. Data Analysis Methods:
The CARS algorithm was used for informative variable selection, and PLS was used for constructing predictive models. Model population analysis (MPA) was conducted to evaluate the influence of training set composition on model performance.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
