矿井突水是采矿生产过程中威胁最大的地质灾害之一，快速有效地判别突水水源是预防矿井水害的关键所在。通过分析潘二煤矿含水层的水化学性质，开展了相邻水岩水锶同位素的测试与分析，选取⁸⁷Sr/⁸⁶Sr、Ca²⁺、Na⁺+K⁺、Mg²⁺、HCO₃⁻、SO₄²⁻、Cl⁻ 7个判别指标，结合主成分分析与Fisher理论、主成分分析与距离判别理论、主成分分析与BP神经网络，分别建立基于锶同位素的混合水源判别模型（Sr-F模型、Sr-D模型、Sr-B模型），利用模型对未知水样进行判识。结果表明：基于锶同位素的Sr-B判识模型的判识效果最好，其准确率达到95%；基于主成分分析与BP神经网络突水水源判别模型能够有效地提高突水水源识别精度，能准确地判识相邻灰岩含水层突水水源，为矿井安全生产提供保障。

Mine sudden water is one of the most threatening geological hazards during mining production, so rapid and effective identification of sudden water sources is the key to prevent mine water damage. In this study, we analyzed the water chemistry of Panji Coal Mine aquifer and carried out the testing and analysis of strontium isotopes of water in the adjacent water rock, selected seven discriminatory indexes: ⁸⁷Sr/⁸⁶Sr, Ca²⁺, Na⁺+K⁺, Mg²⁺, HCO₃⁻, SO₄²⁻, Cl⁻, combined with principal component analysis and Fisher’s theory, principal component analysis and distance discriminatory theory, principal component analysis and BP Neural network, to establish the discriminatory models of mixed water sources based on strontium isotopes (Sr-F model, Sr-D model, Sr-B model), and use the models to discriminate unknown water samples. The results show that the Sr-B model based on strontium isotopes has the best discriminative effect, and its accuracy reaches 95%. Therefore, the identification model of water inrush sources based on principal component analysis and BP neural network can effectively improve the identification accuracy of water inrush sources, accurately identify water inrush sources in adjacent limestone aquifers, and provide guarantee for mine safety production.