建筑结构火灾下倒塌应急监测技术研究

Research on fire-induced collapse emergency monitoring technology for building structures

建筑火灾已经成为威胁城市公众安全的主要灾害之一。为了保护消防救援人员的人身安全,综合运用红外热成像、测量机器人和近景摄影测量等技术构建了一种建筑结构火灾下倒塌应急监测方法,对火灾下建筑的温度场和形变场进行监测,辅助开展建筑倒塌风险分析。该方法采用红外热成像技术获取建筑火灾的温度场数据;利用测量机器人和近景摄影测量技术分别获取关键点位高精度形变数据和多方位立体形变数据;建立了雾凇优化算法(Rime Optimization Algorithm,RIME)与极限学习机(Extreme Learning Machine,ELM)相结合的协同校正模型RIME-ELM,利用关键点位高精度形变数据对近景摄影测量的监测结果进行校正,提高立体形变监测数据的精度。为了验证该方法的有效性,搭建建筑结构实体模型,模拟开展火灾应急监测。结果表明,该方法所获取的火场温度和形变数据符合建筑火灾一般规律,利用RIME-ELM模型所获取的形变校正结果与ELM模型相比,平均相对误差明显降低。这验证了该方法的有效性和可行性,能够为消防应急救援提供全面、可靠的数据支撑。

Building fires have emerged as one of the primary disasters threatening public safety in urban areas.This research establishes an emergency monitoring method for the collapse of building structures during fires by integrating multiple technologies,including infrared thermal imaging,measurement robots,and close-range photogrammetry.The goal is to enhance the safety of civilians and fire rescue personnel.The method was employed to monitor the temperature fields and deformation patterns of buildings during fires,supporting risk analysis of potential collapses and aiding decision-making for on-site fire rescue operations.Specifically,infrared thermal imaging was used to collect temperature field data from the building fires,while measurement robots and close-range photogrammetry were utilized to obtain high-precision deformation data at critical points and multidirectional stereo deformation data of the structures under fire.A calibration model known as RIME-ELM was developed,integrating the Rime Optimization Algorithm(RIME)with the Extreme Learning Machine(ELM).This model optimized the input weight matrix and the bias matrix of the hidden layers,which are initially generated randomly by ELM,through the application of RIME.Based on the model,high-precision key point monitoring data was utilized to refine the monitoring results obtained from close-range photogrammetry,thereby enhancing the accuracy of the stereo deformation data.A solid model of the building structure was constructed to simulate fire emergency monitoring and to validate the effectiveness of this method.The results indicated that both the temperature data and the deformation data obtained through this method aligned with the general principles governing building fires.When compared to the ELM model,the correction results achieved using the RIME-ELM model demonstrated a significant reduction in the mean relative error.It was confirmed that the model is both effective and feasible,providing comprehensive and reliable data support for emergency rescue operations during building fires.