Prediction of the Height of Water Flowing Fractured Zone Based on the MPSO-BP Neural Network Model

Abstract

The height of the water flowing fractured zone (HWFFZ) is the key parameter for safe coal mining under water bodies. In order to improve the accuracy of prediction model of the HWFFZ, the grey relational analysis (GRA) is used to quantitatively analyze the influencing factors of the HWFFZ, the main influencing factors of the HWFFZ under the condition of comprehensive mechanized mining and comprehensive top coal caving mining are selected scientifically, and these five factors that include mining depth, mining thickness, inclined length of working face, hard rock lithology proportion coefficient, and coal seam dip angle are selected as the discriminant indexes for predicting the HWFFZ. Nonlinear inertia weight adjustment, linear adjustments of local acceleration factor and global acceleration factor, and Gaussian mutation operation are introduced into particle swarm optimization (PSO) algorithm to improve the performance of PSO algorithm. The modified particle swarm optimization (MPSO) algorithm is used to optimize the initial weights and thresholds of back propagation (BP) neural network, and the prediction model of the HWFFZ based on MPSO-BP neural network algorithm is established according to the measured data. The prediction results are compared with those of PSO-BP neural network model and empirical formulas, the results show that the prediction accuracy of MPSO-BP neural network prediction model is higher than that of PSO-BP neural network prediction model and empirical formulas, and its prediction mean relative error (MRE) is smallest. The model also has good application effect and certain guiding significance for actual production.