CNN-GRU-Based intelligent online prediction of screw machine rotation speed for Earth pressure balance shield machine

Abstract

Abstract During shield machine construction, the precise determination of control parameters is particularly important for construction safety. By effectively using construction big data and mining the information in it, accurate construction control parameters with geological adaptability can be obtained in real time. To this end, this study proposes an intelligent online prediction method for screw machine rotation speed of Earth pressure balanced (EPB) shield based on convolutional neural network gated recursive unit (CNN-GRU). Firstly, the construction data is processed with Pearson correlation coefficient, the 10-dimensional data of the current moment is selected as the input variable, and screw machine rotation speed at the next moment is chosen as the output. Next, CNN is used to extract features from the input variables to find a nonlinear relationship between the input and output. Then, in order to construct a complete screw machine rotation speed prediction model, the GRU is used to filter the features to establish a more accurate nonlinear relationship. Finally, the control effect of the method was verified by simulation experiments. The simulation results show that the proposed method can accurately predict the rotation speed of the screw machine online. The traditional control mode in shield tunnel construction can lead to unstable excavation surface due to time lag and inaccuracy. Meanwhile, once the excavation surface is unstable, soil collapse or uplift can lead to construction safety problems, while the the required construction data to the control room at the next moment can be quickly and accurately fed by the online prediction method proposed in this study, thereby effectively avoiding the above construction safety problems and having certain practical application value.