Abstract: To address the issues of high feature dimensionality,high computational complexity,empirical dependence in parameter selection,and insufficient generalization ability in mine floor water inrush prediction,a prediction model based on Principal Component Analysis(PCA) and Fruit Fly Optimization Algorithm(FOA) improved Support Vector Classification(SVC),namely the PCA-FOA-SVC model,is proposed. First,PCA is used to reduce the dimensionality of the multi-factor influencing data of floor water inrush. Then,FOA is employed to adaptively and globally optimize the penalty factor and kernel function parameters of SVC. Finally,the optimized SVC is utilized to construct the optimal classification hyperplane for binary classification prediction of floor water inrush conditions. Public data extracted from published academic papers were used as the test set. The test results show that after algorithm optimization,the classification accuracy increases from 79. 8% to 98. 6%,and when the algorithm reaches the optimal fitness value,the iteration number is only 17,with a root mean square error as low as 0. 021. These results verify the effectiveness of PCA and FOA in improving SVC. Further model testing indicates that the prediction accuracy reaches 98. 31%,and the prediction time is only 5. 24 s,significantly outperforming other comparison models. This model can provide effective technical support for early warning and safety prevention of coal mine floor water inrush disasters.