LF ‐ NMR Combined With XGBoost for Nondestructive Prediction of Rice Seed Vigor Under Drying Conditions
Journal of Food Process Engineering · 2026-11 · 49 (11)
DOI: 10.1111/jfpe.70878 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Food Process Engineering, online erschienen am 05.11.2026. Verfasst von Mengnan Yang und 6 weiteren Autor:innen. Die Arbeit nennt 42 Literaturangaben. Gefördert von: National Natural Science Foundation of China.
Abstract (Originalsprache)
Drying must balance moisture removal and rice seed vigor, but the predictive value of internal water‐state features across drying regimes remains unclear. We combined low‐field nuclear magnetic resonance (LF‐NMR) with machine learning to predict vigor index (VI). Six treatments were evaluated in 792 independent experiments with japonica rice “ Sasanishiki .” Treatments comprised drying with and without tempering at 45°C and 55°C, sun‐drying, and shade‐drying. Germination responses were assessed using Welch's analysis of variance, and four regression models were compared. Continuous drying at 55°C produced the highest effective moisture diffusivity (2.03 × 10 −11 m 2 /s). LF‐NMR spectra shifted toward shorter relaxation times, while tempering reduced attenuation of the main bound‐water peak. Tempering increased germination from 86% to 92% and VI from 126.87 to 156.69 at 45°C. At 55°C, germination increased from 42% to 77% and VI from 17.10 to 97.29. Extreme gradient boosting (XGBoost) performed best on the test set ( R 2 = 0.978; RMSE = 8.477). Nested fivefold cross‐validation yielded pooled out‐of‐fold R 2 = 0.967 and RMSE = 10.323. Shapley additive explanations identified strongly bound and bound water peak areas as the most influential predictors. LF‐NMR with XGBoost supports nondestructive vigor assessment, while 45°C tempering offers a favorable balance between drying efficiency and vigor retention among the tested hot‐air treatments.
Förderung
- National Natural Science Foundation of China
Zitieren
APA
Yang, M., Zou, X., Li, W., Shen, Z., Yang, T., Feng, L., & Song, P. (2026). LF ‐ NMR Combined With XGBoost for Nondestructive Prediction of Rice Seed Vigor Under Drying Conditions. Journal of Food Process Engineering, 49(11). https://doi.org/10.1111/jfpe.70878
Vancouver
Yang M, Zou X, Li W, Shen Z, Yang T, Feng L, et al. LF ‐ NMR Combined With XGBoost for Nondestructive Prediction of Rice Seed Vigor Under Drying Conditions. Journal of Food Process Engineering. 2026;49(11). doi:10.1111/jfpe.70878
BibTeX
@article{Yang2026,
author = {Yang, Mengnan and Zou, Xinlin and Li, Wen and Shen, Zhiyi and Yang, Tao and Feng, Longlong and Song, Ping},
title = {LF ‐ NMR Combined With XGBoost for Nondestructive Prediction of Rice Seed Vigor Under Drying Conditions},
journal = {Journal of Food Process Engineering},
year = {2026},
volume = {49},
number = {11},
doi = {10.1111/jfpe.70878}
}
EUVIMED