Surrounding Rock Stability Classification and Application of Coal Roadway Based on GA-BP

In order to get scientific results of surrounding rock stability classification, referring to statistics of 24 classification related papers, the SPSS database of classification index is set up to analyze frequency as well as study the index frequency and distribution characteristics, then the important classification indexes are obtained. Taking the selection principles and the actual situations of Shanxi Coking Coal Group into consider, the eleven classification indexes are determined. By combined use of genetic algorithm and BP neural network, the GA-BP based coal roadway surrounding rock stability classification model is established. The network topology of the model is constructed by GA global heuristic search, and the optimal weight and threshold of the model is determined by GA global optimization and BP local optimization. The model is implemented by MATLAB programming. Eighty coal roadways are chosen as training samples to train the classification ability of the model, then the GA-BP based classification model is applied to twenty coal roadways. The results show that the model accuracy is 95%, with its high nonlinear mapping accuracy, it is suitable for the stability classification of coal roadways in Shanxi Coking Coal Group.