Abstract
Currently, most of the roadways adopt one support design strategy, which leads to high stress and insufficient support parameters in some crushed areas of the roadways and excess support parameters in some stable regions. There is an urgent need for a reliable method of grading the roadway perimeter rock to realize a reasonable support design for the whole area and cycle of the roadways. Taking Xiaobaodang No.1 coal mine as the background, Based on previous research, we utilized SPSS to analyze the data and selected ten indicators that significantly influence roof stability and are easily obtainable. The relatIVe weights between the influencing factors were determined using the hierarchical analysis method. The results showed that fIVe key factors, namely, roadway depth, roof strength, direct roof thickness, mining height, and rock integrity, emphatically affect the roof's stability. Based on the borehole data in the study area of the mine, 40 sets of borehole data were processed using normalization, and based on the weights of the influencing factors, a classification formula for the stability of the roadway perimeter rock was proposed to classify the boreholes initially. The roadway roof stability classification model of the BP neural network is constructed. The accuracy of the training set of 40 sets of drill hole data is 92.8%, and the accuracy of the test set is 91.7%. The classification results of the model are verified by using the mine pressure data of the mined face, and the mine pressure data shows a noticeable step change with the classification results, which puts forward theoretical references for the subsequent differentiated support of the working face. Numerical simulation software is used to analyze the vertical stress of different types of roof layers and the vertical stress of coal pillars, and the vertical stress of coal pillars at the roof layers that are highly prone to collapse of the roof is higher than that at other roof layers, so it is necessary to strengthen the support.