基于U-Net++的建筑物抗震能力评价模型  

Seismic Capacity Evaluation Model for Buildings Based on U-Net++

作　　者：侯红语 窦爱霞[1] 郭红梅[2] HOU Hong-yu;DOU Ai-xia;GUO Hong-mei(Institute of Earthquake Forecasting,China Earthquake Administration,Beijing 100036,China;Sichuan Earthquake Agency,Chengdu 610041,China)

机构地区：[1]中国地震局地震预测研究所,北京100036 [2]四川省地震局,四川成都610041

出　　处：《地震》2024年第4期153-172,共20页Earthquake

基　　金：国家高分专项课题(31-Y30B09-9001-13/15-06);中国地震局地震预测研究所基本科研业务费专项(CEAIEF202403016,CEAIEF20240503)。

摘　　要：针对传统建筑物抗震评估耗时耗力、难以大范围展开的问题,本文提出了一种基于U-Net++的建筑物抗震能力评价模型。该模型以U-Net++为骨干网络,集成了多种信息源,包括高分辨率遥感影像、建筑物建造年代、高度、用途和结构类型。以四川省雅安市雨城区为研究区,设计了17个对比实验探究不同特征对建筑物抗震能力的影响程度,为广域地震灾害风险精细化评估提供了有力的技术支撑。Aiming at the problem that the traditional seismic evaluation of buildings time-consuming and the difficulty in a large-scale implementation,this paper proposes a seismic capacity evaluation model for buildings based on U-Net++.The model takes U-Net++as the backbone network,integrates multiple information sources,such as high-resolution remote sensing images,building construction age,height,use and structural types.Taking the Yucheng District of Ya’an City,Sichuan Province,as the study area,17 comparative experiments were designed to explore the degree of influence that different features have on the seismic resilience of buildings.This work provides robust technical support for refined assessments of earthquake disaster risks on a broad scale.

关 键 词：遥感影像 建筑物 抗震性能 语义分割 U-Net++ 

分 类 号：TP79[自动化与计算机技术—检测技术与自动化装置]