Regional Seismic Damage Prediction of Buildings Based on Bayesian Network and Community Information Database: A Case Study in Shanghai

Abstract

ABSTRACT The development of the Community Information Database (CID) provides an opportunity to enhance the accuracy of predicting regional seismic damage. The CID includes fundamental attributes of buildings and daily operation information such as maintenance record, loading variation, and structural defect inspection. However, existing models cannot leverage the daily operational information of the CID to calibrate the real-time seismic performance of regional buildings. In this study, a novel Bayesian network (BN) model is proposed to assess the regional seismic damage of buildings based on information provided by the CID. Buildings in an area in Shanghai, China, are selected to investigate the applicability and accuracy of the proposed model. The investigation indicated that the proposed model can synthesize seismic fragility curves and matrices for various building types and calibrate the prediction based on daily operational information such as structural defect, structural alteration, and property management. If the CID has missing building attributes, the proposed model can also utilize an embedded subnetwork to supply the database. Therefore, the proposed model integrates more comprehensive information than the traditional methods to enhance the accuracy of predicting the regional seismic damage.