Performance parameter design of seismic isolation bearings for high-speed railway simply-supported bridges using neural network

In the simply-supported bridge system for the high-speed railway with unequal pier heights, the intrinsic relationship between the seismic isolation bearing parameters and the seismic performance objectives of the bridge is very complex. It is not easy to establish a precise mathematical model, and the design process of the bearings relies heavily on the experience of the designers. This paper uses an artificial neural network (ANN) to design the seismic performance of seismic isolation bearing. Considering that the triple friction pendulum bearing (TFPB) can adaptively change the stiffness under different seismic levels, the ANN model fits the mapping relationship between the parameters of the TFPB and the seismic performance objectives. Four evaluation indexes are used to verify that the ANN model has good generalization ability. A case study shows that the designed bearing meets the expected seismic performance requirements, which verifies the applicability of the ANN model.