Abstract: In the coal mine tunnel TBM excavation, in order to identify the broken surrounding rock state in time during the excavation process, a method based on tunnel excavation parameters for identifying the surrounding rock characteristics of coal mine tunnels is proposed. Firstly, according to the abnormal changes in the travel parameters of the support boot hydraulic cylinder during the excavation, it is judged that the TBM excavation has entered the broken surrounding rock section from the stable surrounding rock section, thus obtaining the data set of the stable surrounding rock section and the transitional section. Then, based on the data set, the correlation and recognition ability of the tunneling parameters are analyzed, the excavation parameters for identifying the broken rock characteristics are selected, and finally, a long short-term memory (LSTM) model is used for prediction. According to the prediction results of the excavation parameters, the identification index is obtained, and subsequent identification of the rock breaking state in the following sections is completed. This method is applied in a gas control tunnel of a coal mine, and the precision of the LSTM model in predicting the excavation parameters is higher than 98%. Based on the prediction results, the relative error percentage of the excavation parameters is calculated as an identification index. With a threshold of 5% relative error percentage, the identified rock breaking state exists in three data segments, which is consistent with the judgment of the support boot hydraulic cylinder travel. This shows that this method can effectively identify the broken rock characteristics of the surrounding rock, and has the advantages of high intelligence level and little interference in construction.