This study utilized GC-MS, GC-IMS, E-nose, and E-tongue to analyze the flavor characteristics of Guangchenpi (GCP) from five core producing areas aged 5 to 40 years. Key findings include: W1W, W2S, and W5S sensors in the e-nose and bitter, umami, sweet, sour, and astringent tastes in the e-tongue were pivotal. A total of 219 VOCs were identified, with 43 characteristic VOCs screened out using PLS-DA. The flavor compounds exhibited a 'Λ'-shaped change, peaking at the age of 30 years. Aging had a greater influence on flavor than origin, with Tianma origin (TM) samples exhibiting the strongest flavor. Among seven machine learning prediction models developed, the Random Forest (RF) model showed optimal performance, achieving 100 % accuracy in geographical origin discrimination and 96 % accuracy in aging year prediction. Key characteristic factors including isopropanol and 5-methyl-2(3H)-furanone were validated through the top 10 critical features, confirming model interpretability.