Comparative analysis of BP neural network and RBF neural network in seismic performance evaluation of pier columns

Abstract

With the increasing frequency and intensity of earthquakes, the idea of seismic design continues to develop, and it must be ensured that the structure not only has sufficient strength, but also has certain ductility. However, there is currently no suitable method for effective evaluation of seismic performance. In order to find a method with high accuracy, this paper uses BP neural network algorithm and RBF neural network algorithm to analyze and control the collision response of urban bridge adjacent beam and pier beam under strong earthquake. In this paper, the prediction of concrete carbonation depth and steel corrosion is carried out. The comparison of simulation results shows that the prediction effect of RBF network is better than that of BP network. In this paper, based on the skeleton curve calculated by bending shear failure test column, bending curve, shear deformation and longitudinal reinforcement drawing deformation are considered. The results are in good agreement with the hysteresis curve and can be used to calculate the reinforced concrete column under axial load and horizontal load. As the IM index gradually increases, the damage and failure probability of the bridge structure IDA show a relatively rapid upward trend. Then, when IM approaches 3.5 g, the failure probability of the bridge reaching the corresponding side shift ratio of 1.75% is more than 95%.