A prediction model for the height of the water-conducting fractured zone in the roof of coal mines based on factor analysis and RBF neural network

The development height of the water-conducting fractured zone (WCFZ) is a fundamental parameter related to aquifer protection in coal mines. Accurately predicting the height of the WCFZ is extremely important for preventing and controlling water hazards in the roof of the coal seam and ensuring the safety of coal mining. To accurately predict the development height of the WCFZ in a coal seam roof to better prevent water inrush and protect the ecological environment of the mining area, a prediction model of the development height of the WCFZ was established by using factor analysis (FA) and radial basis function (RBF) neural network based on five influencing factors, including the mining depth, the coal-seam dip angle, the mining height, the uniaxial compressive strength of overburden, and the working-face length. The prediction performance of the model for new sample data was tested and evaluated the mean absolute error, root-mean-square error, and mean relative error. For comparison, a traditional RBF neural network model and a traditional support vector machine (SVM) model were also constructed for prediction analyses. The results show that the FA-RBF neural network model fits the data well. It also has strong generalization ability and good prediction performance for new samples based on the average absolute error, root-mean-square error, and average relative error of 4.47 m, 4.71 m, and 7.52%, respectively, which are better than the traditional RBF neural network prediction and traditional SVM prediction models. The proposed model combines the respective advantages of FA and RBF neural network, and it avoids the disadvantages of traditional prediction methods, which do not consider repeated information interference and noise among influencing factors, and simplify the dimension of the input layer of the neural networks while reducing the scale of the neural networks. The proposed model improves convergence speed, learning ability, and prediction ability. It provides a practical approach for accurately predicting the development height of the WCFZ, which can help prevent and control water inrushes from coal-roof strata and protect the ecological environment of mining areas.