Surrounding rock classification represents distinguishing the different grades of surrounding rock according to the hardness and integrity of surrounding rock. Accurately obtaining the surrounding rock grade of drill jumbo working face is not only the basis for selecting the tunnel position and support type, but also the key to ensure the safety of the drill jumbo's construction site. As the traditional classification methods, engineering drilling and geological mapping are time-consuming and labor-intensive. Aiming at this situation, this paper proposes an intelligent identification method of surrounding rock grade combine drilling parameters with machine learning algorithm XGBoost. Firstly, adequately analyse the correlation between drilling parameters and rock label, and select six drilling parameters as feature vectors for surrounding rock grade recognition. Then outlier processing and data screening are carried out on the data recorded by the drill jumbo. Next, we construct a model based on XGBoost to realize the rapid and accurate identification of surrounding rock grade. Finally, the effectiveness and superiority of the proposed method are demonstrated by the actual data collected by the drill jumbo in Gao Jiaping tunnel, and mix the partial data of Alianqiu tunnel together to construct 5 datasets to compare the identification performance of other classical algorithms. The results show that the recognition capability of the proposed method is superior to those of other algorithms, and the recognition accuracy of surrounding rock along the tunnel can reach 99.68%.

Coal is an important natural resource in China and plays an essential role in the development of industry and national economy. To realize unmanned mining, it is necessary to identify coal-rock type of working face accurately and efficiently. As the photographing is interfered by water mist, dust, air flow, lighting, vibration and other factors, the accuracy of image feature recognition methods are seriously affected. Therefore, this paper proposes an intelligent identification method based on boring parameters of dig windlass and XGBoost algorithm. Firstly, the coupling relationship between machine parameters recorded by dig windlass was analysed to remove a large number of redundant parameters, which reduces 22.7% training time of the model and 63.8% identification time. Secondly, remove the data recorded by the dig windlass under abnormal working conditions. Then, construct a model based on XGBoost algorithm and input the selected parameters and data into the model for training. Finally, the validity of the proposed method is verified by the data collected from the Project of Chai Jiagou Coal Mine in Tongchuan. The results show that the accuracy rate of coal rock type identification is more than 98% even when the training data is very little and the abnormal data under normal working condition is kept, which confirms the effectiveness and strong robustness of the proposed method.