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

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