18 October 2022 in Journal of Food Process Engineering
1 October 2021 in Journal of Materials Science & Technology
25 July 2023 in Journal of Food Process Engineering
1 June 2024 in Journal of Near Infrared Spectroscopy
The rapid determination of moisture content in Panax notoginseng taproot (PNT) was determined using a portable near infrared spectrometer (900∼1700 nm). First, to reduce baseline offset of the spectra Savitzky-Golay and standard normal variate transformation were combined to preprocess the original spectral data. Then, competitive adaptive reweighting sampling and bootstrapping soft shrinkage (BOSS) were employed to extract feature wavelengths that could characterize the moisture content information of PNT respectively. Finally, the least square support vector regression (LSSVR) model was established based on feature spectra and full spectra. To improve the prediction accuracy of the model, a LSSVR model based on the arithmetic optimization algorithm (AOA) was proposed, and the optimization results were compared with those of the snake optimizer and particle swarm optimization. The results indicated that the best prediction model was BOSS-AOA-LSSVR, with r2 and RMSEP values of 0.96 and 0.03%, respectively. Thus, it is feasible to predict the moisture content of Panax notoginseng taproot by portable near infrared spectroscopy in combination with BOSS-AOA-LSSVR. The results show that portable near infrared spectroscopy can be used to predict the moisture content of Panax notoginseng taproot, which provides a theoretical basis for the rapid and non-destructive detection of the moisture content of Panax notoginseng taproots.
5 January 2023 in Journal of Food Safety
1 January 2026 in Journal of Chemistry
To achieve the rapid and nondestructive detection of the content of Panax notoginseng saponins (PNS), samples from Kunming and Wenshan, Yunnan Province, were collected for this study, and their near‐infrared spectral data within the 900–1700 nm range were acquired. Following spectral preprocessing via different methods, the bootstrapping soft shrinkage (BOSS) algorithm was employed to select characteristic wavelengths, and a least squares support vector regression (LSSVR) model was established. To further enhance predictive performance, the equilibrium optimizer (EO), snake optimizer (SO), and arithmetic optimization algorithm (AOA) were introduced for comparative analysis. The results demonstrated that the BOSS–EO–LSSVR model achieved the best predictive performance, with R² = 0.9657, RPDP = 5.4660, RMSEP = 0.2624, and MAEP = 0.2093. In conclusion, near‐infrared spectroscopy can be effectively used to quantitatively determine PNS, and this study provides a solid theoretical foundation for developing rapid detection equipment for Panax notoginseng .