Online Fault Prediction for EPB Shield Tunneling Based on Neural Network
For the complex earth conditions and uncertain factors, the earth pressure balance (EPB) shield tunneling is a complicated and high-risk process. There would cause some faults, such as earth caking, soil occluding in the capsule, water spewing, surface settlement. To avoid them, this paper applies artificial neural network (ANN) to predict the common shielding faults. The neural network is trained by several samples about the tunnel boring machine’s (TBM) parameters, and then it will have self-learning to identify the fault. With the parallel computing ability, the network could detect and predict abnormal behaviors online. This paper includes three parts, firstly, the introduction of EPB, and four usual blockings; and then the principle of BP neural network is present, for the defect of BP algorithm, two kinds of improved BP algorithms are applied in the network; finally, an simulation is given to illustrate the prediction.