摘要
In order to evaluate the seismic fragility of bridges more scientifically and effectively,the paper introduces in detail the procedure and method of the seismic damage prediction of bridges group, based on support vector machine. The paper built a SVM model, by selecting learning samples from earthquake disaster data of 103 bridges in Tangshan earthquake, Haicheng earthaquake and Wenchuan earthquake etc. By comparing the earthquake prediction ability of the SVM model and the BP neural network, we can find that the accuracy of SVM model is higher.The study shows that the SVM model has good generalization capability and can be operated in practice.