Prediction of return oil flow in shield tunneling machine screw conveyors with closed-loop pump controlled motors using an adaptive LSTM-CNN approach

The screw conveyor of the shield tunneling machine is a key component in tunneling process, primarily responsible for transporting cut earth and rock. In this system, the prediction of the return oil flow rate of the closed-loop pump control motor is crucial for adjusting operational parameters and ensuring efficient system operation. Accurate flow predictions can effectively prevent system overloads, thereby avoiding impacts on the efficiency, cost, and safety of the conveying process. This paper presents an innovative prediction model that combines adaptive long short-term memory networks (LSTM) and convolutional neural networks (CNN) to accurately predict the return oil flow rate in the screw conveyor of the shield tunneling machine. The LSTM-CNN model utilizes the temporal dynamics and spatial features inherent in the system's operational data to enhance prediction accuracy. Compared to traditional methods, this model demonstrates superior accuracy and robustness in predicting return oil flow, which is vital for the maintenance and performance prediction of the shield tunneling machine's hydraulic system. The results indicate that the integration of LSTM and CNN significantly improves the predictability of the closed-loop pump control system and optimizes maintenance schedules, thereby reducing downtime and operational costs in shield tunneling projects.