Identification of sulfur-fumigated Shanyao by near-infrared spectroscopy combined with DD-SIMCA

Abstract

Shanyao (Chinese yam) is an important edible biologic medicine, known as the Chinese “little ginseng”, which is extensively sold in vegetable and medicinal material markets. The Shanyao product sold on the market is often sulfur-fumigated for improving its appearance and anti-corrosion and pest resistance effect. It is of great significance to develop a method for distinguishing whether Shanyao is sulfur-fumigated or not. This work explores the feasibility of combining near-infrared (NIR) spectroscopy, pre-processing, feature selection and class-modeling for distinguishing between sulfur-fumigated and non-fumigated Shanyao. Different concentrations of sulfur dioxide and fumigation duration are considered. A total of 200 simulated samples are prepared. Two filtering-like algorithms, i.e., Relief and MRMR, and data-driven soft independent modelling of class analogy (DD-SMICA) are used for spectral feature selection and class-modeling of the target class, respectively. The results indicate the best one-class model can be obtained by a smaller subset of original features on the pre-processing spectra. It is also found that different duration had little effect on the condition of the same concentration of sulfur dioxide. In addition, it seems necessary to perform feature selection, but two methods of feature selection present similar results. It can be concluded that the proposed procedure is a feasible for identifying the sulfur-fumigated Shanyao, which is a reference for quality control of other systems.