The quality and value of tea leaves are significantly influenced by their different storage years, thus establishing a rapid and accurate method for determining storage years is crucial. To explore the feasibility of hyperspectral imaging (HSI) technology for identifying tea storage years, this study used HSI technology in the visible (VIS) and near-infrared (NIR) regions, combined with machine learning (ML) methods, to classify and identify tea samples from six storage years: 2009, 2012, 2015, 2018, 2021, and 2025. Spectral data were collected from the VIS (400 ~ 1000 nm) and NIR (1000 ~ 2400 nm) regions, and spectral preprocessing was performed using first-order derivative (1-st der), second-order derivative (2-nd der), Savitzky-Golay smoothing (SGS), standard normal variable transformation (SNV), and multiplicative scatter correction (MSC). Support vector machine classifier (SVC), random forest classifier (RFC), and K-nearest neighbor classifier (KNNC) models were then constructed for tea storage year classification, and the optimal model was used to perform visualization analysis of the hyperspectral images. The results showed that the MSC-RFC model performed best in the VIS region, with a classification accuracy of 97.22%. In the NIR region, the 1-st der-RFC model demonstrated the best performance, achieving a classification accuracy of 100.00%. The study indicates that HSI technology combined with spectral preprocessing methods and ML models shows certain feasibility for identifying tea storage years based on the sample conditions of this study. However, the generalizability of the method across samples from multiple sources still requires further evaluation through independent external validation.