Virtual modelling based fragility assessment of structures under bushfire propagation
提出一个框架，评估灌木火灾中住宅结构的易损性，并引入扩展支持向量回归（X-SVR）技术实现快速预测，帮助优化结构保护计划。
Driven by increased human activities in rural-urban interfaces, the construction of residential or commercial buildings in these areas is experiencing a notable growing trend. In comparison to those built in urban regions, these structures, constructed in rural-urban interfaces, are in closer proximity to natural vegetation, therefore facing a heightened bushfire risk. The timely execution of Structural Protection Plans (SPP) is of utmost importance in the case of bushfire threats, where a swift response within a short timeframe is necessary, considering the diverse fragility characteristics of structural components. To address this, the present study introduces a novel framework for assessing the fragility of typical residential structures under both low and high wind speed conditions, specifically focusing on three key structural components: window frames, walls, and roofs. The assessment of structural probability-based fragility is performed using the newly developed limit state function and takes into account the influence of multiple non-deterministic factors, including vegetative conditions, wind speed, different temperature thresholds of structural components, and fire response time. Furthermore, to enable rapid prediction of structural probability-based fragility on the fireground, a virtual modelling (VM) technique, named extended support vector regression (X-SVR), is introduced and incorporated into the proposed fragility assessment framework. The efficiency and accuracy of this virtual modelling technique in assessing the bushfire fragility of structures under different wind speed intervals have been investigated and validated through a comprehensive case study of a real Australian house. The proposed framework is poised to provide valuable insights into optimizing SPP by swiftly identifying the most fragile structural components in practice.