Abstract:Achondrites are igneous rocks originating from terrestrial planets or differentiated asteroids, and they preserve crucial insights into planetary differentiation and evolution. However, distinguishing achondrites from terrestrial igneous rocks remains challenging due to their striking similarities. Developing a convenient, non-destructive identification technique is essential for improving the efficiency and accuracy of field meteorite searches and museum curation. This study utilizes 64 meteorite samples analyzed by handheld energy-dispersive X-ray fluorescence (HH-ED-XRF) and 1532 datasets from the literature to train a model that effectively differentiates Vesta, Martian, Lunar, and angrite meteorites from terrestrial rocks. Key parameters, including bulk rock major element compositions, Fe/Mn ratios, and Al?O?/(FeO+MgO) ratios, were used in the training process with Subspace K-Nearest Neighbor algorithm. The model achieved an overall accuracy of 95%. This technique provides intelligent and non-destructive identification of achondrites in the field, significantly enhancing the efficiency and accuracy of meteorite searches. Its applications extend from museum meteorite collection curation to field searches of meteorites in hot and cold deserts.