Abstract: In order to accurately identify the source of mine water inrush, a discriminant model of mine water inrush source based on PCA-ISSA-BP model is proposed. Six conventional ions and PH of different aquifer water in Suntuan Coal Mine were selected as discriminant indexes. Firstly, principal component analysis was used to reduce the dimension of seven indexes. Then, Sine chaotic mapping, fusion sine cosine algorithm and Lévy flight strategy were used to improve SSA to optimize the parameters of BP model. Finally, the model was established by programming on Matlab software. The results show that the principal component analysis reduces the correlation between the indicators and improves the discriminant accuracy of the model. The BP neural network model after normalization and parameter optimization can jump out of the local minimum point, and the convergence speed has been greatly improved. Compared with the ISSA-BP, PCA-SSA-BP, PCA-PSO-BP and PCA-BP models, the PCA-ISSA-BP model has higher recognition accuracy, stronger fitting ability and smaller mean square error. The model can more accurately identify the source of mine water inrush.