基于深度学习的倾斜摄影建筑物表面损毁信息提取

建筑物受损信息是地震受灾程度评估的基础,针对传统建筑物表面信息识别人工成本高、效率低等问题,受深度学习提取建筑物影像的启发,提出利用无人机倾斜摄影模型与深度学习相结合的方法提取震后建筑物表面破损信息。以2019年长宁6.0级地震为例,选用双河镇震后倾斜摄影模型切片图为数据源,对比分析面向对象分类方法、VGG-16模型和DeeplabV3+模型对建筑物表面损毁信息的提取结果。分析结果表明,针对建筑物表面破损信息的提取,尤其是细小裂缝的提取,语义分割网络DeeplabV3+模型具有较强的优势(准确率96.93%、召回率96.85%、总体精度96.89%),可实现建筑物表面破损信息的有效提取,具有较强的应用价值。     

Adaptive Training of Neural Networks for Automatic Seismic Phase Identification

—?A neural network module has been implemented in the Prototype International Data Centre (PIDC) for automated identification of the initial phase type of seismic detections. Initial training of the neural networks for stations of the International Monitoring System (IMS) requires considerable effort. While there are many seismic phases in the analyst-reviewed database that can be assumed as the ground-truth resource of the initial phase type of Teleseism (T), Regional P (P), and Regional S (S), no ground-truth database of noise (N) is available. To reduce analyst effort required in building a ground-truth database, an “Adaptive Training Approach” is proposed in this paper. This approach automatically selects training patterns to take advantage of the learning ability of neural networks and information on the accumulated observation database. Using this approach, neural networks were trained on the data provided by station STKA, Australia. The performance of automated phase identification has been improved significantly by the retrained neural networks. This approach is also validated by comparison with the performance using the ground-truth noise database.     

基于遥感影像的建筑物空间分布方法在地震灾害损失预评估中的应用

房屋建筑数据是地震重点危险区预评估工作的基础,需基于获取到的房屋建筑信息开展人员伤亡、经济损失、救援物资需求等预评估工作。历年地震重点危险区预评估工作能够通过现场调查得到的房屋建筑信息占比极小,仅能进行抽样调查。因此,为批量完成危险区内全部房屋建筑损失估算,需基于遥感影像获取房屋建筑矢量数据,并建立数据库。为实现全国地震重点危险区预评估工作中大批量建筑物矢量化数据的快速获取,本文采用基于遥感影像的建筑物空间分布数据批量获取方法,得到地震重点危险区内建筑物空间矢量数据,结合现场抽样调查得到的建筑物属性信息,建立地震重点危险区建筑物空间分布数据库,进而为地震重点危险区灾害损失预评估工作提供数据基础。本文采用的方法可广泛应用于地震重点危险区房屋建筑信息获取工作中,可提高工作效率,降低工作成本,提升预评估工作的科学性和准确性。     

多尺度分割和深度学习相结合的倾斜摄影三维影像建筑物震害信息提取

采用多尺度分割和深度学习相结合的方法对震后倾斜摄影三维影像建筑物震害信息进行提取,获取建筑物的屋顶和墙体多种破坏信息。以2017年九寨沟M_S7.0地震后倾斜摄影三维影像为例,依据三维影像建筑物顶面和墙体等进行样本的多尺度分割,样本分为完好建筑物面、破坏建筑物面、其它地物和背景等三类,选取211个100×100像素的样本集对卷积神经网络模型进行训练,采用训练后的模型提取灾区千古情风景区和漳扎镇小学的建筑物震害信息,并将提取结果与目视解译结果进行精度对比,结果显示:破坏建筑物面提取精度分别为65.5%和71.1%,总体分类精度分别为82.1%和84.1%,卡帕（Kappa）系数分别为68.7%和64.9%,表明该方法在倾斜摄影三维影像建筑物震害提取方面具有一定的优势。      

Seismic performance assessment of “hybrid” structures using two-level multy group GIS oriented approach: case studies

In this paper, a two-level multi-group approach to seismic performance assessment of existing building stock is presented and discussed. The first level of assessment is a simple visual screening based on the principles originally developed in FEMA (Rapid visual screening of buildings for potential seismic hazards: a handbook—FEMA 154 and 155, Applied Technology Council, Redwood City, 2002). The second level of assessment is based on an integrated approach (field survey, geotechnical investigations, experimental testing and analytical investigations) with incorporated procedures of different levels of complexity. The acceptance criteria for assessment are founded on demand versus capacity relationships, defined at element and story level. The developed approach was applied in performance assessment of the existing buildings in Karposh, which is one of the largest and oldest municipalities in Skopje, the capital of the Republic of Macedonia. Rapid visual screening (level_1 activity) was performed for 161 existing buildings. For two selected buildings, level_2 assessment was carried out. The results from the assessment were uploaded into the existing geo-referenced building inventory database on the Karposh municipality that represents a practical tool for the decision makers in enforcing decisions on design and construction of seismically resilient building environment.     

Classification of Collapsed Buildings for Fast Damage and Loss Assessment

Fast and reliable identification of collapsed buildings is essential in case of earthquake disasters in urban areas. Airborne laserscanning offers the possibility to fulfil this task. Based on height measurements, geometrical surface models of buildings can be generated with this technology. Comparing the undamaged pre-event models with those recorded after an earthquake, the location of collapsed buildings and the dimension and characteristic of their damage can be obtained. The knowledge about typical damage types of collapsed buildings is necessary to interpret the changes found between the pre- and post-event building models. As existing building damage classifications don’t meet the requirements of this novel technique, observations and reports of building collapses were analysed. This leads to a new classification system of collapsed buildings and the definition of the so-called “damage catalogue”.The damage catalogue is a composition of different damage types of entire buildings typically occurring after earthquakes and it contains the observed dimensions of the geometrical features such as volume reduction or inclination change for each damage type. Besides the detectability of these geometrical features in airborne laserscanning data, the differentiation of the damage types takes effects on casualty numbers and on different search and rescue needs into account. The damage catalogue was developed by evaluating the associated database, which contains the characterisation of real damaged buildings by the defined geometrical features.The paper includes the conception of the damage catalogue and of the associated database, their use for the described reconnaissance technique and their further application possibilities.     

基于深度学习网络的地震地质灾害识别研究

遥感影像识别方法是破坏性地震震后地质灾害快速、准确获取的重要方法之一,传统的遥感影像识别方法主要以人工目视识别方法和半自动识别方法为主,需投入大量的人力和时间。针对破坏性地震震后地质灾害解译时间长、投入人力多等问题,以2017年8月8日四川九寨沟7.0级地震震后高分辨率无人机遥感影像为研究样本,提出基于深度学习网络的地震地质灾害识别方法。首先结合震后遥感影像解译资料和现场调查资料,提取九寨沟地震地质灾害无人机遥感影像特征,并构建研究区地震地质灾害解译指标和分类数据集;然后采用DeepLabv3+网络结构及softmax损失函数,建立基于深度学习网络的地震地质灾害遥感影像图像语义分割模型方法;最后采用半监督学习方法进行结果验证。研究结果表明,基于深度学习网络的地震地质灾害识别方法可有效识别九寨沟地震地质灾害分布信息,整体分类识别准确率为94.22%,F₁分数值为0.77,结果具有较好的一致性和准确性,可提升地震现场灾情获取和重点地震隐患识别等工作效率及服务能力。     

Shaking table testing of an existing masonry building: assessment and improvement of the seismic performance

This paper aims to assess and improve the seismic performance of an existing masonry building with flexible floors, representative of a Portuguese building typology—‘gaioleiro’ buildings. The study involved seismic tests and dynamic identification tests of two models (nonstrengthened and strengthened) in the shaking table. Each model was subjected to several seismic tests with increasing amplitude. Before the first test and after each seismic test, the dynamic identification of the model was carried out, aiming at obtaining their seismic vulnerability curves based on a damage indicator obtained from the decrease of the frequencies of the modes. In the strengthened model, steel elements were used to improve the connection between walls and floors, together with ties in the upper stories. The results show that adopted strengthening technique is effective for reducing the seismic vulnerability of ‘gaioleiro’ buildings, namely for improving the out‐of‐plane behavior of the facades. Copyright © 2013 John Wiley & Sons, Ltd.     

Approximate incremental dynamic analysis using the modal pushover analysis procedure

Incremental dynamic analysis (IDA)—a procedure developed for accurate estimation of seismic demand and capacity of structures—requires non‐linear response history analysis of the structure for an ensemble of ground motions, each scaled to many intensity levels, selected to cover the entire range of structural response—all the way from elastic behaviour to global dynamic instability. Recognizing that IDA of practical structures is computationally extremely demanding, an approximate procedure based on the modal pushover analysis procedure is developed. Presented are the IDA curves and limit state capacities for the SAC‐Los Angeles 3‐, 9‐, and 20‐storey buildings computed by the exact and approximate procedures for an ensemble of 20 ground motions. These results demonstrate that the MPA‐based approximate procedure reduces the computational effort by a factor of 30 (for the 9‐storey building), at the same time providing results to a useful degree of accuracy over the entire range of responses—all the way from elastic behaviour to global dynamic instability—provided a proper hysteretic model is selected for modal SDF systems. The accuracy of the approximate procedure does not deteriorate for 9‐ and 20‐storey buildings, although their dynamics is more complex, involving several ‘modes’ of vibration. For all three buildings, the accuracy of the MPA‐based approximate procedure is also satisfactory for estimating the structural capacities for the limit states of immediate occupancy, collapse prevention, and global dynamic instability. Copyright © 2006 John Wiley & Sons, Ltd.     

建筑物震害多源遥感特征与机理分析

随着遥感信息源的不断增加,多种遥感数据被用于详细判读建筑物的震害情况.为准确判读震害等级与建立震害自动识别模式,本文收集整理了汶川地震震区的震害遥感图像,通过目视判读、图像处理、统计分析,重点分析了各类震害建筑物在光学影像中的特征表现、在合成孔径雷达图像中的成像机理特征以及在激光雷达图像中的三维特征.在此基础上构建了建筑物简化模型,并联合光学影像和雷达图像对震害建筑物的影像特征剖面予以分析.结果显示:光学遥感图像色彩信息符合人眼色觉原理,具有较好的直观判读效果;合成孔径雷达图像能够记录地物侧面、表面的粗糙程度和角反射特点,信息量丰富但不直观;激光雷达图像能获取建筑物的三维信息,因此震害评估工作中需有效地综合利用多源遥感数据,才能实现最佳的判识效果.      

Evaluation of seismic vulnerability of a complex RC existing building by linear and nonlinear modeling approaches

Recent earthquakes, that stroked Italian regions in past decades (Umbria— Marche 1997; Molise 2002; L’Aquila 2009), pointed out the high vulnerability of reinforced concrete existing buildings causing severe damages in the structures and consequently life losses. This is mainly due to the fact that such structures were often built without reference to seismic actions or on the basis of old standard provisions. Nowadays in Italy, Public Authorities are requested to evaluate the seismic vulnerability of their building stock assessing the actual capacity of such structures, as a consequence of new hazard levels and seismic microzonation introduced by new standards. According to Eurocode 8 or Italian standard NTC 2008, the seismic analysis of existing reinforced concrete buildings can be performed by one of the established procedure (i.e. Linear Static Analysis LSA, Linear Dynamic Analysis LDA, Nonlinear Static Analysis NSA, Nonlinear Dynamic Analysis NDA), depending on the achieved knowledge level about the structural system and materials. In order to compare efficiency and differences of previously described approaches, a deep investigation was executed on a reinforced concrete existing building whose dynamic behaviour was evaluated by an experimental dynamic analysis. In such a way, updated models were obtained and adopted for seismic analysis performed by using linear and nonlinear approaches, taking into account the stiffness and strength contribution of masonry infill walls. It was so possible to get useful indications on the reliability and discrepancies of different modelling approaches as well as on the influence of masonry infills on the seismic response of existing r.c. buildings.     

Analysis of damage data of low-rise buildings subjected to a shallow Mw6.3 earthquake

In May 2008 a shallow M_w6.3 earthquake struck South Iceland with an epicentre close to two small towns. Nearly 5000 low-rise residential buildings were affected. The recorded maximum PGA was 0.88 g. A great deal of damage occurred, but there was no loss of life. In Iceland all buildings are registered in a detailed official database and insurance against natural disasters is obligatory. As the repair costs for every affected building had to be assessed for insurance purposes this provided an unusual opportunity to review structural performance across the whole population of buildings in the affected area. The estimated repair cost was classified in a number of subcategories covering structural and non-structural damage for five different residential building typologies. Study of these buildings showed that non-structural damage dominated the overall damage. The main monetary damage was cosmetic damage of partition walls and flooring. The structural systems performed quite well and no buildings collapsed.     

A decision-theory methodology for the selection of buildings for strong-motion instrumentation

Current projections indicate that six buildings per year will continue to be instrumented under the California Strong-Motion Instrumentation Program for the next several decades. In order to select these buildings systematically for instrumentation, we have developed a methodology that incorporates the fundamental elements of decision theory. These elements include an identification of the types of buildings that should be instrumented, a definition of the expected severity of ground shaking at each possible building site along with the probability of occurrence, and a quantification of the relative value of obtaining a building-response record for each building type. Using this information, we can apply decision theory to calculate the expected utility (degree of preference) of instrumenting buildings of a particular type at various sites. The sites are then ranked in order of preference for each building type. This procedure, developed for the California Strong-Motion Instrumentation Program, can be extended to instrumentation programs in other areas.     

Spatio-temporal precipitation modelling in rural watersheds

Abstract

The natural variability of precipitation in agricultural regions both in time and space is modelled using extensions of Box & Jenkins (1976) methodology based on the ARMA procedure. This broad class of aggregate regional models belongs to the general family of Space-Time Autoregressive Moving Average (STARMA) processes. The paper develops a three-stage iterative procedure for building a ST ARMA model of multiple precipitation series. The identified model is STMA (1₃). The emphasis is placed on the three stages of the model building procedure, namely identification, parameter estimation and diagnostic checking. In the parameter estimation stage the polytope (or simplex) method and three further classical nonlinear optimization algorithms are used, namely two conjugate gradient methods and a quasi-Newton method. The polytope method has been adopted and the developed model performed well in describing the spatio-temporal characteristics of the multiple precipitation series. Application has been attempted in a rural watershed in southern Canada.     

Probabilistic neural networks for seismic damage mechanisms prediction

The procedure commonly employed to assess the seismic vulnerability of buildings uses simplified qualitative and quantitative observations obtained from the measured data entered into report forms. In Italy, the data sheets adopted by the National Defence Group against Earthquakes (Gruppo Nazionale per la Difesa dai Terremoti—GNDT) play a unifying and reference role. This paper proposes a method for the processing of the data contained in such report forms which is based on probabilistic neural networks producing a Bayesian classification. The final goal is to exploit the fundamental learning and generalization capabilities of neural networks to obtain an estimate of the vulnerability of structural systems. In particular, the aim is to be able to predict the damage mechanisms which may be triggered in the macro-elements of public worship buildings. Copyright © 1999 John Wiley & Sons Ltd.     

Extension of the CSM‐FEMA440 to plan‐asymmetric real building structures

The capacity spectrum method (CSM) has established itself as one of the most used Nonlinear Static Procedures for the seismic assessment of structures, since its introduction in 1975, when it was first presented by Freeman. More recently, this procedure was implemented in the ATC40 guidelines and lately improved in the FEMA‐440 report. The first step of work addressed by this paper relates to the comparison between the two features of the CSM. In the second part, an extension of the FEMA‐440CSM version is proposed for plan‐asymmetric real RC building structures. The case studies under analysis are the SPEAR building—an irregular 3D structure representing typical old three‐storey buildings in the Mediterranean region, from the early 1970s—and two real Turkish buildings with five and eight storeys. The CSM‐ATC40, the CSM‐FEMA440 and the proposed extended CSM‐FEMA440 method are applied and the results obtained duly compared with nonlinear dynamicit timehistory analyses. For the latter, semi‐artificial ground motions are used to define the seismic action. Copyright © 2010 John Wiley & Sons, Ltd.     

盘海营地区农村住宅的抗震性能调查分析

开展农村民居抗震性能调查是国家防震减灾规划的重要组成部分。本文在对辽宁省盘锦、海城、营口地区农村住宅抗震性能调查工作的基础上,从房屋的结构特点、建造年代分布、房屋的大小等方面进行了详细总结,鉴于该区域结构特点和建造年代具有很强的相关性,给出了该调查区域的农村住宅分类方法。提出基于调查中发现的进深与分类之间的相关性,利用高分辨率卫星图片判读能够在保证准确度的情况下有效降低建立农村房屋存量数据库的成本。最后,本文结合震害中农村民居砌体墙破坏模式及《镇（乡）村建筑抗震技术规程》（2008）指出了盘海营地区农村房屋在抗震方面存在的不足。     

Description and analysis of building damage due to Pyrgos, Greece earthquake

On March 1993 an earthquake of magnitude M_s = 5·5 shook Pyrgos, a town in Western Peloponnissos, one of the most seismic prone areas in Greece. The damage induced to modern reinforced concrete buildings was rather light in contrast to the damage induced to historic and traditional buildings of adobe, stone or brick masonry which was severe. In order to study the causes of structural damage, detailed data are collected from a rather large statistical sample of 1023 masonry buildings and 22 reinforced concrete framed buildings with visible damage. For each building the number of storeys, the material of construction, as well as the type and the degree of damage are recorded. In addition, consideration is given to the site of the building within the town and the corresponding soil conditions. For reinforced concrete buildings, damage occured mostly in areas with relatively high estimated spectral accelerations and fundamental soil periods of vibration close to those of the buildings. Nevertheless, further analysis is required to explain the selective damage of a very small number of buildings. For masonry houses, the effect of soil conditions is more systematic. Moreover, the effects of the number of storeys as well as the age and material of construction appear to be dominant.     

基于无人机影像的九寨沟地震建筑物震害定量评估

利用2017年8月8日九寨沟7.0级地震震后获取的无人机影像,结合地面震害调查资料,分析各类建筑物震害特征,建立建筑物震害无人机遥感解译标志;选取地震灾区漳扎镇(部分区域)和荷叶寨2个区域作为研究区,进行了无人机遥感建筑物震害提取,基于遥感震害指数进行了震害定量评估,并与现场建筑物震害调查统计结果进行了比较验证。结果显示,遥感解译建筑物震害与实际震害程度相吻合,表明利用震后快速获取的高分辨率无人机影像,可以较为准确地识别建筑物震害,进而为地震灾害定量评估和应急救援辅助决策提供重要参考。     

Assessing reparability: simple tools for estimation of costs and performance loss of earthquake damaged reinforced concrete buildings

Tools for assessing building reparability via the estimation of expected performance loss (PL) and associated costs for repair of existing RC building classes damaged by an earthquake are presented. The assessment approach relies on the availability of a number of suitably developed: (i) capacity curves for representative building classes; (ii) curves relating global ductility demand μ to the expected PL for the same classes; and (iii) PL–cost for repair relationship calibrated on database collecting cost data of more than 2300 buildings damaged after 2009 L'Aquila earthquake. The tools are developed applying a simplified procedure involving the simulated design of existing building classes, the assumption of predefined collapse mechanism types and the analyses of the seismic behavior of equivalent SDOF systems representative of ‘intact’ and ‘damaged’ structures after an earthquake. The use of these tools may give useful preliminary indications to decision makers for establishing reparability priorities in the aftermath of damaging earthquakes or to insurance companies to value sound insurance premium for existing building classes. Copyright © 2014 John Wiley & Sons, Ltd.