A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
Enhancing earth pressure balance tunnel boring machine performance with support vector regression and particle swarm optimization
Abstract This paper combined data-driven modeling and optimal control for performance enhancement of earth pressure balance tunnel boring machine (EPBM). Two coupled processes, EPBM advance rate (AR) and cutterhead rotation torque, are modeled using support vector regression (SVR). An optimal control framework was formulated to maximize the AR, solved with particle swarm algorithm. Using the Seattle N125 project as case study, it is found the SVR model can predict EPBM AR with R 2 = 0.90, normalized root mean square error (nRMSE) = 0.30 and mean absolute percentage error (MAPE) = 31.2%, and R 2 = 0.65, nRMSE = 0.59 and MAPE = 6.7% for torque prediction. Compared to human operator, EPBM with optimal control can increase AR by 0.6–23.3 mm∕min on average, accompanied by an average torque reduction of 83.1 kN - m. It is found higher cutterhead rotation and lower chamber pressure always contribute to faster tunneling, but the optimal total thrust force to apply depends on the chamber pressure.
Highlights Modeled EPBM tunneling advance rate and cutterhead rotation torque using data-driven method with physical guidance Visualized and interpreted the influence of each EPBM operation to advance rate and torque. Established a general optimal control formulation for EPBM tunneling performance enhancement. Used the Seattle N125 tunneling project as a case study to demonstrate the expected improvement with optimal EPBM control.
Enhancing earth pressure balance tunnel boring machine performance with support vector regression and particle swarm optimization
Abstract This paper combined data-driven modeling and optimal control for performance enhancement of earth pressure balance tunnel boring machine (EPBM). Two coupled processes, EPBM advance rate (AR) and cutterhead rotation torque, are modeled using support vector regression (SVR). An optimal control framework was formulated to maximize the AR, solved with particle swarm algorithm. Using the Seattle N125 project as case study, it is found the SVR model can predict EPBM AR with R 2 = 0.90, normalized root mean square error (nRMSE) = 0.30 and mean absolute percentage error (MAPE) = 31.2%, and R 2 = 0.65, nRMSE = 0.59 and MAPE = 6.7% for torque prediction. Compared to human operator, EPBM with optimal control can increase AR by 0.6–23.3 mm∕min on average, accompanied by an average torque reduction of 83.1 kN - m. It is found higher cutterhead rotation and lower chamber pressure always contribute to faster tunneling, but the optimal total thrust force to apply depends on the chamber pressure.
Highlights Modeled EPBM tunneling advance rate and cutterhead rotation torque using data-driven method with physical guidance Visualized and interpreted the influence of each EPBM operation to advance rate and torque. Established a general optimal control formulation for EPBM tunneling performance enhancement. Used the Seattle N125 tunneling project as a case study to demonstrate the expected improvement with optimal EPBM control.
Enhancing earth pressure balance tunnel boring machine performance with support vector regression and particle swarm optimization
Yu, Hongjie (author) / Zhou, Xu (author) / Zhang, Xiaoli (author) / Mooney, Michael (author)
2022-06-23
Article (Journal)
Electronic Resource
English
Argentina's first earth pressure balance tunnel boring machine
| British Library Conference Proceedings | 1998
| British Library Online Contents | 2017