Structural Control and Health Monitoring · 2018 · 17 citations · 34 references
EngineeringCivil EngineeringConstruction EngineeringOptimizationTunnel EngineeringTunnel Boring Machine DesignTunnel Boring Machine OperationTunnel MonitoringTunnel SafetyTunnel Settlement ControlTunnelingConstruction Parameter OptimizationParticle Swarm ControlParticle Swarm OptimizationRadial Basis Function NetworksSupport Vector ClassificationSupport Vector Machine ClassificationTuningTunnel Boring Machines
Settlement is a key indicator in underground engineering, and monitoring it during tunneling beneath an operational subway tunnel is essential to keep face and grout pressures within a specific range. The study proposes a hybrid approach that uses uniform design and a radial basis function neural network to model the relationship between settlement and influencing factors. The resulting model serves as a tuning module for the tunneling boring machine, while a total‑station‑based soft‑computing framework employing a particle‑swarm‑optimized support vector machine forecasts settlement of the rail. An illustrative case on Changsha Metro Line 3 demonstrates that the metamodel accurately predicts settlement, confirming its feasibility.
The settlement is regarded as an important index in underground engineering. When tunneling under across the operational subway tunnel, settlement of the operational tunnel should be monitored. To control the tunneling-induced movement of the operational subway tunnel, the operation parameters of mechanized-tunneling, namely, face and grout pressures, should be kept in a specific range. A hybrid approach is proposed utilizing uniform design method, and radial basis function neural network to develop the relation of settlement and related influential factors. Such connection is used as a tuning module for tunneling boring machine (TBM). Furthermore, implementing total station robot and soft computing method of support vector machine (SVM), a forecast model for settlement of the rail is established. The prediction tool SVM is improved by using the particle swarm optimization. Parameters c and g from the SVM and γ in the kernel function of the SVM are optimized using the particle swarm optimization. An illustrative case in Changsha Metro Line 3 constructing under the Metro Line 1 tunnel validates the prediction model. The feasibility of the metamodel is demonstrated by means of the example.
34
Genetic algorithms in search, optimization, and machine learning
Choice Reviews Online · 1989 · 49.3K citations
Corinna Cortes, Vladimir Vapnik · Machine Learning · 1995 · 33K citations · Full text
Uniform Design: Theory and Application
Kai-Tai Fang, Dennis K. J. Lin, Peter Winker et al. · Technometrics · 2000 · 883 citations
A generalized discrepancy and quadrature error bound
Fred J. Hickernell · Mathematics of Computation · 1998 · 715 citations · Full text
Training and Testing Low-degree Polynomial Data Mappings via Linear SVM
Yin-Wen Chang, Cho‐Jui Hsieh, Kai‐Wei Chang et al. · 2010 · 478 citations