Identification and prediction method for acoustic emission and electromagnetic radiation signals of rock burst based on deep learning

With the deep development of underground rock engineering, the threat of rock burst disasters is increasing. At present, the identification and prediction of rock burst mostly rely on the experience of field staff to determine the critical value and development trend, and there is a lack of efficient and intelligent methods for the utilization of massive data. Therefore, this paper constructs a rock burst signal recognition and prediction model based on deep learning methods to solve the above problems. In this paper, the acoustic emission (AE) and electromagnetic radiation (EMR) data of the site are first marked and input into the long-short-term memory-fully connected neural network model to realize the identification of rock burst danger signals. Then, the graph data of the AE and EMR sensor monitoring networks are constructed and input into the spatiotemporal graph convolutional network signal prediction model to predict future monitoring data. Finally, this paper uses the same dataset to compare and analyze several other commonly used deep learning models. The results show that the model constructed in this paper has the best performance in the identification and prediction of AE and EMR signals with rockburst risk. This study can provide theoretical reference for intelligent monitoring and early warning of rock burst in underground rock engineering.