Abstract: With the increase of coal mining depth and intensity, mine water accidents occur frequently, and rapid and accurate identification of water sources has become the key to prevent and control water hazards. An improved Fisher discriminant algorithm is proposed to address the problems that the traditional Fisher discriminant model relies on fixed ion combinations and lacks adaptive feature selection mechanism. The algorithm can automatically select the key discriminant according to the hydrological conditions of the mining area by introducing a dynamic feature selection mechanism, which effectively solves the problem of lack of adaptability caused by relying on the combinations of fixed ions in the traditional method, and avoids the interference of redundant information. Taking Tunlan Mine as the research object, the model was constructed based on 19 hydro-chemical discriminants, and the optimal combination of 8 factors was determined by automatic screening, and the performance of the model was verified by combining the back generation estimation method. The results show that the improved model significantly improves the discrimination flexibility and accuracy, and the total accuracy of Tunlan Mine water inflow discrimination can reach 92.97%, which is about 22% higher than that of the traditional method. The model is jointly tested with water sample data from multiple mining areas, and the model shows significant universality, and its performance in the validation of multiple mining areas verifies its adaptive advantage. Meanwhile, the three-dimensional discriminative spatial display analysis further reveals the distribution pattern and mixing characteristics of different water sources, which provides a basis for the subsequent optimization of misjudged samples. The algorithm proposed in this study has high efficiency, adaptability and strong interpretability, and the model can be used to quickly and accurately identify the water sources of water inflow in the future, which can provide key decision support for the prevention and control of water hazards in coal mines, and reduce the economic losses and safety risks caused by water hazard accidents.