研究目的
Exploring the feasibility of combining near-infrared (NIR) spectroscopy and three kind of data descriptions for distinguishing cashmere with other animal ?bers.
研究成果
NIR spectroscopy combined with relief and data description is feasible for identifying different categories of animal fiber. The SVDD model exhibits the best performance with the most flexible and tightest boundary.
研究不足
The study requires more samples to produce more reliable results. The performance of SVDD heavily depends on the width of the Gaussian kernel.
1：Experimental Design and Method Selection:
The study combines NIR spectroscopy with three data description methods (SVDD, KNNDD, GAUSS) for identifying different animal fibers. The Relie? algorithm is used for variable selection and PCA for exploratory analysis and feature extraction.
2：Sample Selection and Data Sources:
A total of 395 samples belonging to four categories (cashmere, wool, rabbit hair, camel hair) were collected from local supermarkets in China.
3：List of Experimental Equipment and Materials:
NIR spectra were collected on the Antaris II FT-NIR spectrophotometer, equipped with an integrating sphere and a standard sample accessory (Thermo Electron Co., USA).
4：Experimental Procedures and Operational Workflow:
NIR measurements were performed in the range of 4000–10,000 cm?1 with 32 scans at a resolution of 3.856 cm?1. Spectral pre-processing involved Savitsky-Golay filter-based first derivative.
5：856 cm?Spectral pre-processing involved Savitsky-Golay filter-based first derivative.
Data Analysis Methods:
5. Data Analysis Methods: The performance of the data description methods was evaluated based on their ability to correctly identify target samples and outliers.
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
