Study on Prediction of Rock Burst in Tunnel Construction Based on BP Neural Network

The factors affecting rock burst,through comprehensive analysis and study on the previous criteria thereof,were divided into three classes: character of surrounding rock,initial geostress and excavation disturbance;then four indices reflecting the above factors were selected as the parameters for prediction,carrying out the multi-factor forecasting for rock burst.By means of BP neural network,the abilities of large-scale information processing with strong robustness and fault-tolerance are available,which are capable of solving the troubles in difficult expression of relationship between tunnel rock burst and various affecting factors as well as improper distribution of the weight of each factor so that the prediction of rock burst in the future based on the prior cases could be implemented.As regards the calculation of prediction model,it may be carried out by use of the perfect mathematical software-Matlab,and the neural network toolbox.The comparison of the calculating result with the constructing practice demonstrates that this method and the model are feasible and effective in rock burst prediction.