Resilience assessment of regional areas against earthquakes using multi-source information fusion

Abstract

This paper proposes a hybrid information fusion approach that integrates the Dempster-Shafer (D-S) evidence theory and cluster analysis, in order to assess the resilience of regional areas against earthquakes at an early stage. The developed resilience assessment framework involves indicators within geological, building, and social dimensions. Basic probability assignments (BPAs) of evidence are determined based on mass functions (MFs) firstly. Subsequently, these BPAs are fused by a new evidence fusion method incorporating the K-means clustering. The proposed evidence fusion method comprises two steps, namely intra-evidence fusion and inter-evidence fusion, to mitigate the adverse effects of conflicts in the multi-source evidence. Defuzzification is implemented on the fused BPA to obtain a representative value for the quantitative evaluation of seismic resilience. A global sensitivity analysis (GSA) is employed to investigate the impacts of various indicators on the resilience of regional areas. A case study in Nepal is used to examine the practicability and validity of the proposed approach. In the case study, the seismic resilience of Nepal at the district level is assessed in a quantitative manner. The research results imply that: (1) There are 84.4% of districts that possess a relatively low overall seismic resilience within the country of Nepal. (2) Building foundation type, internal wall type, and population density are the most sensitive factors contributing to the degree of resilience. (3) The proposed approach tends to provide more rational assessment results with conflicting evidence, compared with the conventional fusion method. The developed approach can be used as a decision tool to estimate regional resilience and provide insights into proactive control and damage mitigation.