为了解决矿井涌水预测问题，引入深度学习理论，将长短期记忆网络（LSTM）和门控循环单元（GRU）进行结合，选取矿井涌水量为研究对象，建立一种LSTM-GRU的矿井涌水预测模型。以陕西某矿的矿井涌水量为样本数据，采用7∶3的比例将数据集划分为训练集和测试集，选择模型训练效果较好的梯度下降算法确定网络模型参数和正则化参数，为了证明LSTM-GRU模型的预测精度，同时将结果分别与传统的ARIMA模型和LSTM模型预测矿井涌水所得到的预测结果进行对比。结果表明：LSTM-GRU复合模型的平均绝对百分比误差（RMSE）为70.51，均方根误差（MAE）为53.4，平均绝对误差（MAPE）为2.80%，可决系数（R²）为0.86，具有较高的预测精度和可靠性，预测效果优于传统的ARIMA模型和LSTM模型。

In order to solve the problem of mine water surge prediction, we introduce deep learning theory, combine long short-term memory network (LSTM) and gated circulation unit (GRU), select mine water surge as the research object, and establish a mine water surge prediction model based on LSTM-GRU. Taking the mine water inflow of a mine in Shaanxi Province as sample data, the data set was divided into a training set and a test set using a 7∶3 ratio, and the gradient descent algorithm with good model training effect was selected to determine the network model parameters and regularization parameters. In order to prove the prediction accuracy of the LSTM-GRU model, the prediction results were compared with those obtained by the traditional ARIMA model and the LSTM model to predict mine water gusher, respectively. The results show that: the mean absolute percentage error (RMSE), root mean square error (MAE), mean absolute error (MAPE) and coefficient of determination (R²) of the LSTM-GRU composite model are 70.51, 53.4, 2.80% and 0.86, indicating that the model has high prediction accuracy and reliability. The prediction effect is better than the traditional ARIMA model and LSTM model.