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

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