基于媒体词云与自然语言处理的城市建筑震损评估

孔庆钊; 计柯妍; 熊冰; 周伯昌; 熊青松; 袁程

doi:10.6052/j.issn.1000-4750.2022.08.0684

基于媒体词云与自然语言处理的城市建筑震损评估

孔庆钊^1,,
计柯妍^1,,
熊冰^1,,
周伯昌^{2, 3, ,},
熊青松^1,,
袁程^1,

1.
同济大学土木工程防灾国家重点实验室，上海 200092
2.
上海市地震局，上海 200062
3.
上海佘山地球物理国家野外科学观测研究站，上海200062

基金项目: 国家自然科学基金项目(2020YFC1512504，52108470)；上海余山地球物理国家野外科学观测研究站项目(SSOP202104)

详细信息

作者简介:
孔庆钊(1988−)，男，上海人，教授，博士，博导，主要从事结构健康监测与数字孪生研究(E-mail: qkong@tongji.edu.cn)

计柯妍(1999−)，女，云南人，博士生，主要从事结构健康监测与智能感知评估研究(E-mail: 2310052@tongji.edu.cn)

熊　冰(1995−)，男，贵州人，博士生，主要从事结构健康监测与人工智能研究(E-mail: bingxiong@tongji.edu.cn)

熊青松(1997−)，男，湖北人，博士生，主要从事结构健康监测与智能感知评估研究(E-mail: 2011158@tongji.edu.cn)

袁　程(1989−)，男，山东人，博士，主要从事结构加固与智能运维研究(E-mail: 20310146@tongji.edu.cn)

通讯作者:
周伯昌(1982−)，女，湖南人，高工，硕士，主要从事地震工程与结构抗震研究(E-mail: zhoubochang1022@126.com)

中图分类号: TP183；P315.63
计量
- 文章访问数: 111
- HTML全文浏览量: 31
- PDF下载量: 43
出版历程
- 收稿日期: 2022-08-04
- 修回日期: 2023-09-27
- 网络出版日期: 2024-03-18
- 刊出日期: 2024-10-24

SEISMIC DAMAGE ASSESSMENT OF URBAN BUILDING STRUCTURES BASE ON ONLINE MEDIA WORLD CLOUD AND NATURAL LANGUAGE PROCESSING

1.
State Key Laboratory for Disaster Reduction in Civil Engineering, Tongji University, Shanghai 200092, China
2.
Shanghai Earthquake Agency, Shanghai 200062, China
3.
Shanghai Sheshan National Geophysical Observatory, Shanghai 200062, China

摘要

摘要:
地震作用下基础设施工程的损毁会造成区域功能瘫痪，对城市区域的震损评估与救援及重建工作提供重要参考。为了做出快速明智的决策并对重大灾害事件做出快速响应，需要对建筑损坏的空间分布和严重程度进行快速评估。考虑到地震灾情后受灾群体在网络媒体和社交平台的信息要新、要快于专家进行实地评估的速度，而受灾区域的建筑物破坏状态有可能从轻微破坏到完全倒塌，对每一栋建筑逐栋细致评估是复杂且耗时的过程。为此，该研究试图利用自然语言深度学习模型BERT对基于震损破坏的文本描述对建筑物损伤进行分类评估。使用《中国地震烈度表》(GB/T 17742−1999)对受灾区域的单个建筑物的破坏状态进行预分类。为了验证所提方法的有效性，该研究收集了2016年台南地区6.4级地震后的300多座单体建筑物并建立了文本词云数据集。该数据集由343个建筑物(121个轻微破坏标签，128个中度破坏标签，94个严重破坏标签)组成，其中85%作为训练与验证数据集，剩余的15%用于鲁棒性测试。当BERT模型识别测试集的GB/T 17742−1999标签时，整体准确率达到 81%。该研究只是试图建立一种新的高效震损评估方法，使用该方法对媒体与社交平台的震损文本描述进行分类评估，能够为应急救援决策提供信息依据并节省一定的战略部署时间。
- 震损评估 /
- 自然语言处理 /
- 数据驱动 /
- 机器学习 /
- 抗震性能
Abstract:
The damage of infrastructure projects under the action of earthquakes may result in regional function paralysis, and the earthquake damage assessment of urban areas can provide important information for rescue and reconstruction work. To make quick and informed decisions and respond quickly to major disaster events, a rapid assessment of the spatial distribution and severity of building damage is required. To this end, this study attempts to use the natural language deep learning model BERT to classify and evaluate building damage based on textual descriptions of earthquake damage. GB/T 17742−1999 is used to pre-classify the damage status of individual buildings in the affected area. To verify the effectiveness of the proposed method, this study collects the data of more than 300 individual buildings after the 2016 M6.4 earthquake in Tainan and establishes a text word cloud database. The dataset consists of 343 buildings (121 labeled light damage, 128 labeled moderate damage, 94 labeled severe damage), of which 85% are used as training and validation datasets, and the remaining 15% are used for robustness testing. When the BERT model recognizes the GB/T 17742−1999 labels of the test set, the overall accuracy rate reaches 81%. This study is just an attempt to establish a new and efficient earthquake damage assessment method. This method can be used to classify and evaluate the earthquake damage text descriptions on media and social platforms, and provide information for emergency rescue decisions and save a certain amount of time for strategic deployment.
- seismic damage assessment /
- NLP /
- data-driven /
- machine learning /
- seismic performance

HTML全文

图 1 地震后的建筑破坏

Figure 1. Building damage after earthquakes

下载: 全尺寸图片幻灯片

图 2 震损等级分布

Figure 2. Distribution of damage index

下载: 全尺寸图片幻灯片

图 3 不同破坏等级的高频词云库

Figure 3. High-frequency word cloud of damage levels

下载: 全尺寸图片幻灯片

图 4 BERT深度网络架构

Figure 4. Pipeline of BERT

下载: 全尺寸图片幻灯片

图 5 BERT数据处理机制

Figure 5. BERT data processing mechanism

下载: 全尺寸图片幻灯片

图 6 评估指标对比

Figure 6. Comparison of evaluation indicators

下载: 全尺寸图片幻灯片

图 7 不同模型的混淆矩阵

Figure 7. Confusion matrix for different models

下载: 全尺寸图片幻灯片

图 8 震损即时评估WEB应用程序

Figure 8. WEB application for seismic instant assessment

下载: 全尺寸图片幻灯片

表 1 建筑破坏等级及震损指数

Table 1 Building damage level and seismic damage index

破坏等级	震损指数	数据集标签	数量
轻微破坏 Light damage	$0 {\leqslant} d < 0.30$	0	121
中度破坏 Moderate damage	$0.30 {\leqslant} d < 0.85$	1	128
严重破坏 Severe damage	$0.85 {\leqslant} d < 1$	2	94

下载: 导出CSV

def create_model(bert_model, max_len = MAX_LEN): ##参数### opt = tf.keras.optimizers.Adam (learning_rate = 1e-5, decay = 1e-7) loss = tf.keras.losses.CategoricalCrossentropy() accuracy = tf.keras.metrics.CategoricalAccuracy() input_ids = tf.keras.Input (shape = (max_len,), dtype = 'int32') attention_masks = tf.keras.Input (shape = (max_len,), dtype = 'int32') embeddings = bert_model ([input_ids,attention_masks])[1] output = tf.keras.layers.Dense (3, activation = "softmax") (embeddings) model = tf.keras.models.Model (inputs = [input_ids, attention_masks], outputs = output) model.compile (opt, loss = loss, metrics = accuracy) return model

下载: 导出CSV

表 2 模型参数汇总

Table 2 Summary of model parameters

输入	输出	超参	连接
Input_1	[None, 128]	0	[]
Input_2	[None, 128]	0	[]
TFBertModel	Last_hidden_state= [None,128,768]; Pooler_output= [None, 768]	109482240	input_1[0][0] input_2[0][0]
Dense	[None, 3]	2307	TFBertModel[0][1]
Total params: 109,484,547
Trainable params: 109,484,547
Non-trainable params: 0

下载: 导出CSV

表 3 BERT模型训练与验证精度

Table 3 BERT model training and validation accuracy

震损等级	Precision	Recall	f1-score	Support
0-轻微	1.00	0.69	0.82	13
1-中度	0.67	0.83	0.74	12
2-严重	0.83	0.91	0.87	11
Micro avg	0.81	0.81	0.81	36
Macro avg	0.83	0.81	0.81	36
Weighted avg	0.84	0.81	0.81	36
Samples avg	0.81	0.81	0.81	36
注：MMicro avg为微平均；Macro avg为宏平均；Weighted avg为加权平均；Samples avg为样本平均。

下载: 导出CSV

参考文献(27)

[1]	陆新征, 张万开, 柳国环. 基于推覆分析的RC框架地震倒塌易损性预测[J]. 地震工程与工程振动, 2012, 32(4): 1 − 6. LU Xinzheng, ZHANG Wankai, LIU Guohuan. Prediction of seismic collapse vulnerability of RC frame based on pushover analysis method [J]. Earthquake Engineering and Engineering Vibration, 2012, 32(4): 1 − 6. (in Chinese)
[2]	陆新征, 唐代远, 叶列平, 等. 我国7度设防等跨RC框架抗地震倒塌能力研究[J]. 地震工程与工程振动, 2011, 31(5): 13 − 20. LU Xinzheng, TANG Daiyuan, YE Lieping, et al. Study on the seismic collapse resistance of RC frame structures with equal spans in zones with seismic intensity Ⅶ [J]. Earthquake Engineering and Engineering Vibration, 2011, 31(5): 13 − 20. (in Chinese)
[3]	徐积刚, 吴刚, 汤昱川, 等. 近场地震下混凝土框架考虑不确定性的抗震风险与恢复性评估[J]. 土木工程学报, 2019, 52(10): 46 − 55. XU Jigang, WU Gang, TANG Yuchuan, et al. Seismic risk and resilience assessment of reinforced concrete frame structures under near fault earthquakes considering uncertainties [J]. China Civil Engineering Journal, 2019, 52(10): 46 − 55. (in Chinese)
[4]	张令心, 江近仁, 刘洁平. 多层住宅砖房的地震易损性分析[J]. 地震工程与工程振动, 2002, 22(1): 49 − 55. ZHANG Lingxin, JIANG Jinren, LIU Jieping. Seismic vulnerability analysis of multistory dwelling brick buildings [J]. Earthquake Engineering and Engineering Vibration, 2002, 22(1): 49 − 55. (in Chinese)
[5]	康现栋, 付皓然, 赵光, 等. 单体建筑抗震韧性评估方法研究与应用[J]. 土木工程学报, 2021, 54(8): 37 − 42. KANG Xiandong, FU Haoran, ZHAO Guang, et al. Research and application of evaluation methods for earthquake resilience of individual building structure [J]. China Civil Engineering Journal, 2021, 54(8): 37 − 42. (in Chinese)
[6]	LU X Z, GUAN H. Earthquake disaster simulation of typical urban areas [M]// LU X Z, GUAN H. Earthquake Disaster Simulation of Civil Infrastructures. Singapore: Springer, 2021: 877 − 932.
[7]	陆新征, 曾翔, 许镇, 等. 建设地震韧性城市所面临的挑战[J]. 城市与减灾, 2017(4): 29 − 34. LU Xinzheng, ZENG Xiang, XU Zhen, et al. Challenges in building earthquake-resilient cities [J]. City and Disaster Reduction, 2017(4): 29 − 34. (in Chinese)
[8]	方东平, 李全旺, 李楠, 等. 社区地震安全韧性评估系统及应用示范[J]. 工程力学, 2020, 37(10): 28 − 44. doi: 10.6052/j.issn.1000-4750.2019.11.0670 FANG Dongping, LI Quanwang, LI Nan, et al. An evaluation system for community seismic resilience and its application in a typical community [J]. Engineering Mechanics, 2020, 37(10): 28 − 44. (in Chinese) doi: 10.6052/j.issn.1000-4750.2019.11.0670
[9]	宗成才, 冀昆, 温瑞智, 等. 城市燃气管网三维度抗震韧性定量评估方法[J]. 工程力学, 2021, 38(2): 146 − 156. doi: 10.6052/j.issn.1000-4750.2020.04.0219 ZONG Chengcai, JI Kun, WEN Ruizhi, et al. Three-dimensional seismic resilience quantification framework for the urban gas network [J]. Engineering Mechanics, 2021, 38(2): 146 − 156. (in Chinese) doi: 10.6052/j.issn.1000-4750.2020.04.0219
[10]	李钢, 薛璐, 余丁浩, 等. 考虑交通阻滞的区域建筑震后功能恢复能力评估[J]. 土木工程学报, 2021, 54(12): 1 − 11. LI Gang, XUE Lu, YU Dinghao, et al. Three-dimensional seismic resilience quantification framework for the urban gas network [J]. China Civil Engineering Journal, 2021, 54(12): 1 − 11. (in Chinese)
[11]	陆新征, 田源, 许镇, 等. 一种新型城市地震灾害模拟方法: 城市抗震弹塑性分析方法[J]. 科学通报, 2020, 65(11): 1055 − 1062. doi: 10.1360/TB-2019-0679 LU Xinzheng, TIAN Yuan, XU Zhen, et al. A new urban earthquake disaster simulation method: City-scale nonlinear time-history analysis [J]. Chinese Science Bulletin, 2020, 65(11): 1055 − 1062. (in Chinese) doi: 10.1360/TB-2019-0679
[12]	谢礼立. 城市防震减灾能力的定义及评估方法[J]. 地震工程与工程振动, 2006, 26(3): 1 − 10. XIE Lili. A method for evaluating cities' ability of reducing earthquake disasters [J]. Earthquake Engineering and Engineering Vibration, 2006, 26(3): 1 − 10. (in Chinese)
[13]	张风华, 谢礼立, 范立础. 非工程因素对城市地震灾害损失影响评价方法[J]. 同济大学学报(自然科学版), 2004, 32(7): 872 − 877. ZHANG Fenghua, XIE Lili, FAN Lichu. Method for evaluating effect of non-engineering factors on city's earthquake disaster loss [J]. Journal of Tongji University (Natural Science), 2004, 32(7): 872 − 877. (in Chinese)
[14]	张娟, 张明媛, 李钢. 地震灾害作用下老旧建筑群韧性评估与提升[J]. 土木工程学报, 2022, 55(1): 41 − 52. ZHANG Juan, ZHANG Mingyuan, LI Gang. Resilience assessment and improvement of old buildings under earthquake disaster [J]. China Civil Engineering Journal, 2022, 55(1): 41 − 52. (in Chinese)
[15]	卢啸. 钢筋混凝土框架核心筒结构地震韧性评价[J]. 建筑结构学报, 2021, 42(5): 55 − 63. LU Xiao. Seismic resilience evaluation of reinforced concrete frame core tube structure [J]. Journal of Building Structures, 2021, 42(5): 55 − 63. (in Chinese)
[16]	吕西林, 陈云, 毛苑君. 结构抗震设计的新概念——可恢复功能结构[J]. 同济大学学报 (自然科学版), 2011, 39(7): 941 − 948. LYU Xilin, CHEN Yun, MAO Yuanjun. New concept of structural seismic design: Earthquake resilient structures [J]. Journal of Tongji University (Natural Science), 2011, 39(7): 941 − 948. (in Chinese)
[17]	何政, 安宁, 徐菁菁. 考虑损伤的结构抗震可恢复性[J]. 工程力学, 2017, 34(5): 179 − 187. doi: 10.6052/j.issn.1000-4750.2015.12.0975 HE Zheng, AN Ning, XU Jingjing. Structural seismic resilience considering damage [J]. Engineering Mechanics, 2017, 34(5): 179 − 187. (in Chinese) doi: 10.6052/j.issn.1000-4750.2015.12.0975
[18]	LIN P H, WANG N Y. Building portfolio fragility functions to support scalable community resilience assessment [J]. Sustainable and Resilient Infrastructure, 2016, 1(3/4): 108 − 122.
[19]	ZHANG W L, LIN P H, WANG N Y, et al. Probabilistic prediction of postdisaster functionality loss of community building portfolios considering utility disruptions [J]. Journal of Structural Engineering, 2018, 144(4): 04018015. doi: 10.1061/(ASCE)ST.1943-541X.0001984
[20]	LIN P H, WANG N Y. Stochastic post-disaster functionality recovery of community building portfolios I: Modeling [J]. Structural Safety, 2017, 69: 96 − 105. doi: 10.1016/j.strusafe.2017.05.002
[21]	LIN P H, WANG N Y. Stochastic post-disaster functionality recovery of community building portfolios II: Application [J]. Structural Safety, 2017, 69: 106 − 117. doi: 10.1016/j.strusafe.2017.05.004
[22]	LIN P, ROSENHEIM N P, WANG N, et al. A stochastic approach to model household re-occupancy in a community following a natural hazard [C]// 13th International Conference on Applications of Statistics and Probability in Civil Engineering. Seoul: ICASP, 2019: 1 − 7.
[23]	鲍跃全, 陈智成, 魏世银, 等. 结构健康监测数据科学与工程研究进展[J]. 工程(英文), 2019, 5(2): 234 − 242. doi: 10.1016/j.eng.2018.11.027 BAO Yuequan, CHEN Zhicheng, WEI Shiyin, et al. The state of the art of data science and engineering in structural health monitoring [J]. Engineering, 2019, 5(2): 234 − 242. doi: 10.1016/j.eng.2018.11.027
[24]	鲍跃全, 李惠. 人工智能时代的土木工程[J]. 土木工程学报, 2019, 52(5): 1 − 11. BAO Yuequan, LI Hui. Artificial Intelligence for civil engineering [J]. China Civil Engineering Journal, 2019, 52(5): 1 − 11. (in Chinese)
[25]	张书颖, 陈适之, 韩万水, 等. 基于集成学习的FRP加固混凝土梁抗弯承载力预测研究[J]. 工程力学, 2022, 39(8): 245 − 256. doi: 10.6052/j.issn.1000-4750.2021.06.0422 ZHANG Shuying, CHEN Shizhi, HAN Wanshui, et al. Study on prediction of frp strengthened reinforced concrete beam’s moment bearing capacity based on ensemble learning algorithm [J]. Engineering Mechanics, 2022, 39(8): 245 − 256. (in Chinese) doi: 10.6052/j.issn.1000-4750.2021.06.0422
[26]	喻泽成, 李启明, 谢龙隆, 等. 基于穷举搜索策略与逻辑回归算法的RC柱地震破坏模式判别模型[J]. 工程力学, 2022, 39(10): 99 − 110. doi: 10.6052/j.issn.1000-4750.2021.06.0429 YU Zecheng, LI Qiming, XIE Longlong, et al. Discrimination model of seismic failure mode of rc columns based on exhaustive search strategy and logistic regression algorithm [J]. Engineering Mechanics, 2022, 39(10): 99 − 110. (in Chinese) doi: 10.6052/j.issn.1000-4750.2021.06.0429
[27]	DEVLIN J, CHANG M W, LEE K, et al. BERT: Pre-training of deep bidirectional transformers for language understanding [C]// Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers) Minneapolis: Association for Computational Linguistics, 2019: 4171 − 4186.