古建筑火灾风险智能孪生评估方法研究

Research on fire risk assessment method of ancient buildings based on intelligent twin modeling

为实现有效的古建筑灾害预防和应急处置决策,采用“数据-处理-业务-服务”的多层次框架,与贝叶斯网络和K-Means聚类算法(BN-KM算法),基于数字孪生模型建立1种古建筑(群)火灾风险评估及决策建议的智能方法,通过8地古建筑火灾历史案例验证该模型的多区域适用性(准确率为87.50%),通过G古建筑(群)4个时段的火灾风险演化规律,验证该方法的时间动态适用性。研究结果表明:该方法中多层次框架可实现多源异构数据(历史数据、监测数据等)的集成量化和风险评估的秒级运算,孪生方法可在事故发生或未造成重要损失前进行计算推演得到高风险区域,并针对不同对象(决策者、救援者、公众)提供专业服务,本文描述新智能方法的评估和决策建议流程,该方法论可拓展应用至其他灾害场景(地震等)。研究结果可为古建筑灾害预防和应急处置提供意见支持。

To achieve the effective disaster prevention and emergency disposal decision-making of ancient buildings,the multi-level architecture of“data-processing-business-service”and the BN-KM algorithm(Bayesian network and K-Means clustering)were adopted,and an intelligent method for the fire risk assessment and decision-making suggestion of ancient buildings based on the digital twin model was established.The multi-regional applicability of this method was verified through eight historical cases of ancient building fires,and the accuracy was 87.50%.The time-dynamic applicability of this method was verified through the fire risk evolution law of G ancient building(group)in four time periods.The results show that the multi-level architecture can realize the integration and quantification of heterogeneous data from multiple sources(e.g.historical data,monitoring data,etc.),and the second-by-second computation of fire risk assessment.The digital twin method can calculate and deduce the high-risk areas before the accident occurs or without causing important losses.Meanwhile,it can provide professional services for different objects(decision-makers,rescuers,and the public).The research results describe the assessment and decision-making suggestion process of the new intelligent method.This methodology may be applied to other disaster scenarios(earthquake,etc.),and provide opinion support for disaster prevention and emergency disposal of ancient buildings.