Abstract: To investigate the hydrochemical characteristics, evolution, and assessment of water quality in the northern foot of the Luoyang Mountains in the western Sichuan Daliang Mountains, 15 sets of local groundwater chemistry samples from different sections of the study area were collected as research objects. Analysis and study of hydrochemical characteristics and evolution in this area use Gibbs’ diagram and ion proportion coefficient method. Furthermore, assessing groundwater quality BP neural network classification method with RMSprop algorithm, supporting services to help local communities develop and use water resources wisely and safely. The results show that the water chemistry of the study area is dominated by Mg²⁺·Ca²⁺−HCO₃⁻. The hydrochemical evolution of groundwater in this area are mainly derived from rock weathering-dissolution. It is controlled by rocks of silicate and carbonate, with silicate playing a more significant role. Considering the geological background of this area, silicate mainly comes from volcanic, clastic, granitic, sandstone and mud stone. The BP neural network was used to train 5000 groups of groundwater samples, and the samples in the study area were evaluated. The training image of the model showed that the BP neural network could well fit the training set of groundwater samples and accurately judge the test set. The result of groundwater quality in this area indicates that Class I water quality points accounted for 13.3%, Class II water quality points accounted for 40%, Class III water quality points accounted for 46.6%. Its overall water quality is good, and the Class III water quality area needs to strengthen groundwater pollution source investigation as well as water quality protection.