Abstract
Gastrodia elata Blume (G. elata) is highly favored in the edible sector owing to its rich nutritional content and distinct flavor. Herein, headspace solid phase microextraction gas chromatography-mass spectrometry (HS-SPME-GC-MS) and Fourier transform infrared spectroscopy (FTIR) technology were employed to classify the origin of G. elata and quantify volatile organic compounds (VOCs). GC-MS revealed that sweet, fruity, and nutty are the key flavor characteristics of G. elata, with samples from Zhaotong City, Yunnan Province, exhibiting superior flavor and richness Based on FTIR data, the gray wolf optimizer-support vector machine and residual convolutional neural network achieved 100 % accuracy in G. elata traceability, with an F(1) of 1.000. Additionally, the partial least squares regression model successfully quantified the main components 2-Nonenal and 2(3H)-Furanone, dihydro-5-propyl- in G. elata, with prediction set residual deviations of 2.6003 and 2.3883, respectively. This approach offers a novel framework for monitoring VOCs quality control in other foods.