基于优选地震强度参数的地下倒虹吸结构易损性分析OACSTPCD

Seismic fragility analysis has proven to be an effective way to study the seismic performance of structures.Compared with above-ground hydraulic structures such as ferries and dams,the seismic vulnerability of underground inverted siphon structures had been less studied.Taking the underground inverted siphon structure in Xiazhuang,Yunnan Province as the research object,16 scalar ground vibration intensity measures(IM)are selected and soil-structure dynamic nonlinear finite element calculations are carried out for the underground inverted siphon structure using the IDA method.Four parameters of effectiveness,practicality,efficiency and relevance are used as evaluation indexes to preferably select scalar IM to form vector IMs.A GA-BP neural network is established by optimizing a back propagation(BP)neural network using genetic algorithm(GA),and the preferable vector IMs are used as input for training.The vulnerability surface of the underground inverted siphon structure is established by the trained neural network.The results show that the preferentially composed vector IMs can better respond to the seismic performance of the underground inverted siphon structure,and can reduce the computational cost of seismic susceptibility analysis by 70% for the structure in this paper.