Optimization design of stability of coal mine roadway support structure based on improved GA-BP model

To improve the stability of coal mine roadway support structures, Huainan Coal Mine was studied as the experimental object, and a Genetic Algorithm (GA) is proposed to optimize the Backpropagation (BP) neural network model. The global optimization and local prediction are coordinated through adaptive coding and other methods. GA is employed to optimize BP weights and thresholds to construct a loosening circle prediction model, and FLAC3D and orthogonal experiments are utilized to optimize support parameters. The experiment shows that the proposed GA-BP model has high accuracy in predicting loose circles. In the optimization of support parameters, the optimal combinations of support parameters for the development roadway, preparation roadway, and mining passage were determined, and the roof settlement under each working condition was controlled within the ideal range, resulting in significant support effects. The stability optimization design method for coal mine roadway support structure proposed in the study can accurately predict the extent of surrounding rock loosening zone and optimize support parameters, providing new ideas and methods for the stability design of deep roadway.