In this paper, we propose a method for classifying tea quality levels based on near-infrared spectroscopy. Firstly, the absorbance spectra of Huangshan Maofeng tea samples were obtained in a wavenumber range of 10,000~4000 cm⁻¹ using near-infrared spectroscopy. The spectral data were then converted to transmittance and smoothed using the Savitzky–Golay (SG) algorithm. The denoised transmittance spectra were dimensionally reduced using principal component analysis (PCA). The characteristic variables obtained using PCA were used as the input variables and the tea level was used as the output to establish a support vector machine (SVM) classification model. The penalty factor c and the kernel function parameter g in the SVM model were optimized using particle swarm optimization (PSO) and comprehensive-learning particle swarm optimization (CLPSO) algorithms. The final experimental results show that the CLPSO-SVM method had the best classification performance, and the classification accuracy reached 99.17%.

In this study, a paper-based microfluidic surface-enhanced Raman spectroscopy (SERS) device was employed for the determination of trace level thiram. The paper-based microfluidic device was fabricated by cutting a hydrophilic region which had been printed on the filter paper and then pasting it onto sellotape. The Au@Ag nanoparticles (NPs) were synthesized with a 30 nm Au core and 7 nm Ag shell and used as the SERS probe. The synthesized nanoparticles were dropped in one of the sample adding zones of the paper-based microfluidics and the thiram solution was dropped in another one. The solutions flowed through their own channels by capillary action and mixed together in the reaction chamber. The optimization studies on the use of paper-based microfluidic devices are discussed. In SERS measurements, the intensity of the peak at 1143 cm⁻¹ was highly sensitive, and so it was chosen as an ideal peak for the quantitative analysis of the concentration of thiram solution. The limit of detection (LOD) of thiram was as low as 1.0 × 10⁻⁹ mol L⁻¹, and the relative standard deviation (RSD) results analyzed at 10 random spots in the SERS measurement area were all below 10%. The recovery values of thiram in adulterated tea samples were from 95% to 110%. All these results suggest that this proposed method is a prospective candidate for trace level thiram detection.

This work provides a novel approach to monitor the aflatoxin B₁ (AFB₁) content in maize by near-infrared (NIR) spectra-based deep learning models that integrates Markov transition field (MTF) image coding and a convolutional neural network (CNN) strategy. According to the data structure characteristics of near-infrared spectra, new structures of one-dimensional CNN (1D-CNN) and two-dimensional MTF-CNN (2D-MTF-CNN) were designed to construct a deep learning model for the monitoring of AFB₁ in maize. The results obtained showed that compared with the 1D-CNN model, the performance of the 2D-MTF-CNN model had been significantly improved, and its root mean square error of prediction, coefficient of predictive determination, and relative percent deviation were 1.3591 μg·kg⁻¹, 0.9955, and 14.9386, respectively. The results indicate that the MTF is an effective data encoding technique for converting one-dimensional spectra into two-dimensional images. It more intuitively reflects the intrinsic characteristics of the NIR spectra from a new perspective and provides richer spectral information for the construction of deep learning models, which can ensure the detection accuracy and generalization performance of deep learning quantitative detection models. This study provides a new analytical perspective for the chemometrics analysis of the NIR spectroscopy.