Huangqi is an important medicinal herb, specifically the dried root of either Astragalus membranaceus (Fisch.) Bge. var. Mongholicus (Bge.) Hsiao (A. mongholicus) or Astragalus membranaceus (Fisch.) Bge. Generally, simulated wild cultivation of A. mongholicus tends to result in higher active compound content compared to horizontal cultivation. However, these differences are challenging to distinguish by visual inspection alone. Traditional methods like High-Performance Liquid Chromatography (HPLC) are accurate for measuring these compounds but are costly and time-consuming. This study aims to develop a rapid, cost-effective, and accurate method to differentiate between simulated wild and horizontally cultivated A. mongholicus. Using a spectroradiometer, we measured the active compound content in ground root samples using HPLC and obtained hyperspectral reflectance data within the 350～2 500 nm wavelength range (SVC-HR1024). The study focused on the spectral characteristics in the visible (VIS, 350～700 nm), near-infrared (NIR, 700～1 100 nm), and shortwave infrared (SWIR, 1 100～2 500 nm) regions. Four machine learning models—Random Forest (RF), K-Nearest Neighbors (KNN), Naive Bayes (NB), and Support Vector Machine (SVM)—were employed for classification. An importance analysis of SHAP features was conducted on the best-performing RF model. The findings reveal that: (1) Simulated wild cultivation had significantly higher active compound content in A. mongholicus than horizontal cultivation; (2) Distinct spectral differences exist between simulated wild and horizontally cultivated A. mongholicus in the NIR and SWIR regions, indicating the impact of the diverse simulated wild environment on pigment synthesis, tissue structure, and chemical composition; (3) The RF model achieved the best performance with an accuracy, precision, F1 score, Kappa, and MCC coefficients of 97.14%, 97.42%, 0.971 3, 0.942 9, and 0.945 6, respectively; (4) SHAP analysis identified key wavelengths associated with moisture, protein, and cellulose content. This study demonstrates the effectiveness of hyperspectral reflectance in distinguishing between simulated wild and horizontally cultivated A. mongholicus samples, providing a novel, non-destructive, and rapid detection method for the quality control and identification of medicinal herbs. This approach has the potential to play a significant role in the quality assessment and market regulation of medicinal herbs.