Structural Modal Parameter Recognition and Related Damage Identification Methods under Environmental Excitations: A Review

Modal parameters can accurately characterize the structural dynamic properties and assess the physical state of the structure.Therefore,it is particularly significant to identify the structural modal parameters according to the monitoring data information in the structural health monitoring(SHM)system,so as to provide a scientific basis for structural damage identification and dynamic model modification.In view of this,this paper reviews methods for identifying structural modal parameters under environmental excitation and briefly describes how to identify structural damages based on the derived modal parameters.The paper primarily introduces data-driven modal parameter recognition methods(e.g.,time-domain,frequency-domain,and time-frequency-domain methods,etc.),briefly describes damage identification methods based on the variations of modal parameters(e.g.,natural frequency,modal shapes,and curvature modal shapes,etc.)and modal validation methods(e.g.,Stability Diagram and Modal Assurance Criterion,etc.).The current status of the application of artificial intelligence(AI)methods in the direction of modal parameter recognition and damage identification is further discussed.Based on the pre-vious analysis,the main development trends of structural modal parameter recognition and damage identification methods are given to provide scientific references for the optimized design and functional upgrading of SHM systems.

Fusion of color and hallucinated depth features for enhanced multimodal deep learning-based damage segmentation

Recent advances in computer vision and deep learning have shown that the fusion of depth information can significantly enhance the performance of RGB-based damage detection and segmentation models.However,alongside the advantages,depth-sensing also presents many practical challenges.For instance,the depth sensors impose an additional payload burden on the robotic inspection platforms limiting the operation time and increasing the inspection cost.Additionally,some lidar-based depth sensors have poor outdoor performance due to sunlight contamination during the daytime.In this context,this study investigates the feasibility of abolishing depth-sensing at test time without compromising the segmentation performance.An autonomous damage segmentation framework is developed,based on recent advancements in vision-based multi-modal sensing such as modality hallucination(MH)and monocular depth estimation(MDE),which require depth data only during the model training.At the time of deployment,depth data becomes expendable as it can be simulated from the corresponding RGB frames.This makes it possible to reap the benefits of depth fusion without any depth perception per se.This study explored two different depth encoding techniques and three different fusion strategies in addition to a baseline RGB-based model.The proposed approach is validated on computer-generated RGB-D data of reinforced concrete buildings subjected to seismic damage.It was observed that the surrogate techniques can increase the segmentation IoU by up to 20.1%with a negligible increase in the computation cost.Overall,this study is believed to make a positive contribution to enhancing the resilience of critical civil infrastructure.

DAMAGE DETECTION IN BUILDINGS USING A TWO-STAGE SENSITIVITY-BASED METHOD FROM MODAL TEST DATA

Many multi-story or highrise buildings consisting of a number of identical stories are usually considered as periodic spring-mass systems. The general expressions of natural frequencies, mode shapes, slopes and curvatures of mode shapes of the periodic spring-mass system by utilizing the periodic structure theory are derived in this paper. The sensitivities of these mode parameters with respect to structural damages, which do not depend on the physical parameters of the original structures, are obtained. Based on the sensitivity analysis of these mode parameters, a two-stage method is proposed to localize and quantify damages of multi-story or highrise buildings. The slopes and curvatures of mode shapes, which are highly sensitive to local damages, are used to localize the damages. Subsequently, the limited measured natural frequencies, which have a better accuracy than the other mode parameters, are used to quantify the extent of damages within the potential damaged locations. The experimental results of a 3-story experimental building demonstrate that the single or multiple damages of buildings, either slight or severe, can be correctly localized by using only the slope or curvature of mode shape in one of the lower modes, in which the change of natural frequency is the largest, and can be accurately quantified by the limited measured natural frequencies with noise pollution.

A MODEL IDENTIFICATION METHOD OF VIBRATING STRUCTURES FROM INCOMPLETE MODAL INFORMATION

The accurate mathematical models for complicated structures are verydifficult to construct.The work presented here provides an identification method for estimating the mass, damping , and stiffness matrices of linear dynamical systems from incompleteexperimental data. The mass, stiffness, and damping matrices are assumed to be real,symmetric, and positive definite. The partial set of experimental complex eigenvalues and corresponding eigenvectors are given. In the proposed method the least squaresalgorithm is combined with the iteration technique to determine systems identified matrices and corresponding design parameters. several illustrative examples, are presented to demonstrate the reliability of the proposed method .It is emphasized thatthe mass, damping and stiffness martices can be identified simultaneously.

A MODEL IDENTIFICATION METHOD OF VIBRATING STRUCTURES FROM INCOMPLETE MODAL INFORMATION

The accurate mathematical models for complicated structures are very difficult to construct.The work presented here provides an identification method for estimating the mass.damping,and stiffness matrices of linear dynamical systems from incomplete experimental data.The mass,stiffness and damping matrices are assumed to be real,symmetric,and positive definite The partial set of experimental complex eigenvalues and corresponding eigenvectors are given.In the proposed method the least squares algorithm is combined with the iteration technique to determine systems identified matrices and corresponding design parameters.Seeveral illustative examples,are presented to demonstrate the reliability of the proposed method .It is emphasized that the mass,damping and stiffness matrices can be identified simultaneously.

提高棉／Modal混纺生条质量的途径

从棉与莫代尔纤维混纺的工艺路线入手，主要介绍了开清棉、梳棉生产工序、对工序半制成品的质量进行了测试，介绍了提高生条质量的主要途径。

Damage Identification under Incomplete Mode Shape Data Using Optimization Technique Based on Generalized Flexibility Matrix

A generalized flexibility–based objective function utilized for structure damage identification is constructed for solving the constrained nonlinear least squares optimized problem. To begin with, the generalized flexibility matrix (GFM) proposed to solve the damage identification problem is recalled and a modal expansion method is introduced. Next, the objective function for iterative optimization process based on the GFM is formulated, and the Trust-Region algorithm is utilized to obtain the solution of the optimization problem for multiple damage cases. And then for computing the objective function gradient, the sensitivity analysis regarding design variables is derived. In addition, due to the spatial incompleteness, the influence of stiffness reduction and incomplete modal measurement data is discussed by means of two numerical examples with several damage cases. Finally, based on the computational results, it is evident that the presented approach provides good validity and reliability for the large and complicated engineering structures.

桥梁健康监测数据的质量评估方法研究

桥梁健康监测数据的挖掘和分析工作只有在整体数据质量符合基本要求的有效数据基础上进行,才能保障如模态参数识别、损伤识别和状态评估等后续工作的准确性。因此,基于量化改进的探索性分析方法(Exploratory Data Analysis,EDA)和相关性分析从数据完整性、准确性和一致性的角度建立了桥梁健康监测静、动态数据的质量评估方法。对某大跨度斜拉桥健康监测系统的静、动态数据进行质量评估,通过对比分析了不同评估质量的温度数据、静挠度数据和不同评估质量的主梁竖向加速度动力信号的模态参数识别的稳定图,验证了所提方法的正确性。结果表明,所提评估方法能够快速有效地判断数据质量的好坏,进而确保桥梁结构的服役性能评估和预测的准确性,有利于提高健康监测数据的可用性和效能。

大数据环境下多模态融合的大学生异常行为预警

针对“校园大数据”累积的海量数据呈现出离散性、稀疏性等问题,如何从基数大、活动广、个性强的校园学生群体中检测出潜在的、有异常行为的学生,已成为学生异常行为分析亟需解决的问题.本文提出了一种大数据环境下基于多模态融合的大学生异常行为预警方法(early warning method for abnormal behavior of college students based on multi-modal fusion in big data environment,EWMAB).首先,针对学生行为画像的表征不够丰富,行为标签存在时效性、动态性等问题,建立一种基于多模态特征深度学习的跨模态学生行为画像模型;其次,针对学生异常行为预测、预警的时效性和后置性问题,在学生行为画像和学生行为分类预测基础上,提出了一种基于多模态融合的学生异常行为预警方法,通过长短期记忆神经网络(long and short term memory networks,LSTM),结合学生行为多指标数据和文本信息来解决学生异常行为预警问题;最后,本文通过应用实例验证模型以学生学习成绩异常预警为例,与其他预警算法相比,EWMAB方法可以提高预警的准确性,实现学生异常行为预警的时效性和前置性,从而使学生教育工作更具有针对性、个性化和预测性.

基于多模态神经网络的微地震事件检测

针对微地震有效信号时序特征存在的局限导致微地震事件识别准确率不高的问题,提出了一种基于多模态学习的神经网络微地震事件检测方法。首先,利用道集数据的相关性以目标道为轴对称制作多道时域模态,对目标道进行时频分析得到S域模态特征;然后,联合时域模态和S域模态设计微地震事件检测神经网络,综合多模态的特征进行训练学习,提高微地震事件识别的精度;最后,为验证方法的有效性,对合成微地震信号进行低信噪比数据分析、小幅值数据分析以及实际油井微地震监测信号事件分析。结果表明,该方法可以有效检测低信噪比及微弱的微地震事件;与支持向量机、卷积神经网络、基于监督机器学习方法的对比实验结果表明该方法具有更高的抗噪性与准确率。

复杂场景下多模态点云数据配准技术

针对复杂环境下多模态点云数据获取难,以及对点云数据配准、三维模型构建精度的要求越来越高的情况。本文以南通大剧院实景三维建模为例,当初始点云和校准点云两组多模态融合点云位置差较大时,采用ICP算法进行点云配准易导致局部最优问题,利用所提出的基于控制点辅助约束的最近点迭代(CPA-ICP)算法通过对点云数据进行配准,并与其他3种点云配准算法的试验进行对比,可知该方法的配准精度和配准效率较高,对复杂场景下的多模态点云数据融合有较好的参考意义。

基于POD/DMD降阶模型的离心泵蜗壳内非稳态流动分析

为深入研究离心泵蜗壳内非线性耦合流动的本质流动特征,本文以一单级单吸式离心泵蜗壳为研究对象,分别采用本征正交分解和动态模态分解2种降阶模型对蜗壳内非定常流场数据进行模态分析,获取流场主导流动结构,并对原流场进行重构。研究发现:蜗壳内流结构主要由速度正负交错,并且周期性特征与叶频及其倍频相关的成对涡旋组成。本征正交分解和动态模态分解方法均可以捕捉流场主要流动结构,并对原流场进行准确还原,两者重构流场与原流场的均方根误差均在0.4%以内。尽管本征正交分解方法在重构流场时整体均方根误差更小,但无法获取单频率流动结构,而动态模态分解方法可以获得不同频率流动结构对流场的贡献,从而捕捉到复杂流场中的不稳定模态。研究成果可以为增强离心泵全局流动的认知、关键水力部件优化设计及发展主/被动流动控制提供理论参考。

基于可泛化模型的复杂商务场景数据分析

随着全球经济合作与竞争形势的快速发展,复杂商务场景日益涌现。其中,数据的规模和复杂性也呈现显著增长,在带来可利用数据资源的同时,也为有效挖掘和分析潜在商务模式带来挑战。本文首先针对复杂商务场景下的数据特点进行分析,其次调研了商务分析方法的研究进展,包括:大规模商务数据集、多模态预训练与特征融合方法以及多场景下可泛化商务模型技术。此外,对复杂商务场景数据分析在未来的应用发展进行思考与总结,讨论了目前亟需解决的关键问题,包括:对更多模态的适应性、商务分析模型可解释性、动态环境下的分布变化鲁棒性等。进一步分析了复杂商务场景可泛化模型的研究现状和挑战,旨在为复杂商务分析领域的研究提供初步参考,实现复杂商务场景的全面智能化。

面向垂直行业定制的多模态网络编程技术

多模态网络(Polymorphic Network,PINet)支持多种用户服务在统一的物理环境中共存,实现用户定制的多样化网络协议和能力等。文章以增量式、全维度、一体化和高安全的网络编程为目标,提出一种多模态网络的编程系统方案和运行逻辑,探讨关键技术包括网络模态生成、安全功能验证和流水线柔性编译等,为各种网络模态的灵活、快速部署提供一种可行的方案。

多模态数据融合的加工作业动态手势识别方法

为了解决单模态数据所提供的特征信息缺乏而导致的识别准确率难以提高、模型鲁棒性较低等问题,提出了面向人机交互的加工作业多模态数据融合动态手势识别策略。首先,采用C3D网络模型并在视频的空间维度和时间维度对深度图像和彩色图像两种模态数据进行特征提取;其次,将两种模态数据识别结果在决策层按最大值规则进行融合,同时,将原模型使用的Relu激活函数替换为Mish激活函数优化梯度特性;最后,通过3组对比实验得到6种动态手势的平均识别准确率为96.8%。结果表明:所提方法实现了加工作业中动态手势识别的高准确率和高鲁棒性的目标,对人机交互技术在实际生产场景中的应用起到推动作用。

四流输入引导的特征互补可见光-红外行人重识别

目前可见光-红外行人重识别研究侧重于通过注意力机制提取模态共享显著性特征来最小化模态差异。然而,这类方法仅关注行人最显著特征,无法充分利用模态信息。针对此问题,本文提出了一种四流输入引导的特征互补网络(QFCNet)。首先在模态特定特征提取阶段设计了四流特征提取和融合模块,通过增加两流输入,缓解模态间颜色差异,丰富模态的语义信息,进一步促进多维特征融合;其次设计了一个次显著特征互补模块,通过反转操作补充全局特征中被注意力机制忽略的行人细节信息,强化行人鉴别性特征。在SYSU-MM01,Reg DB两个公开数据集上的实验数据表明了此方法的先进性,其中在SYSU-MM01的全搜索模式中rank-1和m AP值达到了76.12%和71.51%。

多源异构数据融合关键技术与政务大数据治理体系

随着信息技术的飞速发展,各级政府和大型企业掌握的数据量正在以指数级别增长。然而,数据来源多样会导致格式差异,数据质量参差不齐会影响应用效果,数据分散管理会弱化关联汇集,数据形态异构会造成语义鸿沟。在此背景下,多源异构数据融合负责将来源不同的多模态数据进行有效整合,完成数据互补与关联,进而实现信息增强。目前,大多数已有研究的关注重点集中在大数据治理流程与多模态深度学习,很少有工作研究讨论完整的多源异构数据融合技术框架。因此,在综述关键技术的基础上,文中提出了一整套涵盖“数据引接-数据清洗-数据集成-数据融合”全过程的多源异构数据融合关键技术框架,并对各个环节需要解决的问题与重点任务进行介绍。然后,通过一个政务应用实例场景,给出了政务大数据治理体系的设计,以解决政务数据来源广泛、质量参差不齐、管理分散、形态异构的问题,并进一步阐述了多源异构数据融合的重要价值。最后总结全文并展望未来。

融合多模态数据的小样本命名实体识别方法

作为自然语言处理领域的关键子任务,命名实体识别通过提取文本中的关键信息,帮助机器翻译、文本生成、知识图谱构建以及多模态数据融合等许多下游任务深度理解文本蕴含的复杂语义信息,有效地完成任务.在实际生活中,由于时间和人力等成本问题,命名实体识别任务常常受限于标注样本的稀缺.尽管基于文本的小样本命名实体识别方法已取得较好的泛化表现,但由于样本量有限,使得模型能提取的语义信息也十分受限,进而导致模型预测效果依然不佳.针对标注样本稀缺给基于文本的小样本命名实体识别方法带来的挑战,提出了一种融合多模态数据的小样本命名实体识别模型,借助多模态数据提供额外语义信息,帮助模型提升预测效果,进而可以有效提升多模态数据融合、建模效果.该方法将图像信息转化为文本信息作为辅助模态信息,有效地解决了由文本与图像蕴含语义信息粒度不一致导致的模态对齐效果不佳的问题.为了有效地考虑实体识别中的标签依赖关系,使用CRF框架并使用最先进的元学习方法分别作为发射模块和转移模块.为了缓解辅助模态中的噪声样本对模型的负面影响,提出一种基于元学习的通用去噪网络.该去噪网络在数据量十分有限的情况下,依然可以有效地评估辅助模态中不同样本的差异性以及衡量样本对模型的有益程度.最后,在真实的单模态和多模态数据集上进行了大量的实验.实验结果验证了该方法的预测F1值比基准方法至少提升了10%,并具有良好的泛化性.

基于变分模态分解和复合变量选取的短期负荷预测

精准的短期负荷预测是实现电网精益化运行和管理重要保障,但存在短期负荷波动性强、负荷预测关键影响因素选取困难等精准预测难题。利用变分模态分解将原始电力负荷数据分解为多个子序列,挖掘短期负荷波动特征的同时避免模态混叠问题,提出复合变量选取算法分析筛选影响负荷波动的关键因素,有效去除预测干扰信息并进一步简化预测模型的复杂度,通过兼顾数据短期依赖和长期依赖的长短时记忆神经网络对各子序列进行预测,并将各子序列预测结果进行叠加实现最终的短期负荷预测,据此建立基于变分模态分解和复合变量选取的短期负荷预测方法。选取2019年整年长沙市实际数据验证结果表明,提出算法在复杂外部影响因素下,能准确筛选负荷预测的关键影响因素,相比传统预测模型,提出模型结构更简单、预测精度更高。

基于多模态大数据的国家安全风险态势感知模型构建

[研究目的]为强化国家安全情报能力,推动风险监测预警能力提升,构建基于多模态大数据的国家安全风险态势感知模型。[研究方法]该文对国家安全风险态势感知进行阐释,并分析多模态大数据驱动下的国家安全风险态势感知运作逻辑;在此基础上,构建态势感知模型并解析其内涵。[研究结论]该模型集国家安全风险态势察觉、态势理解、态势预测及态势投射为一体,其构建可为强化国家安全风险监测预警能力提供参考,为服务国家治理体系和能力现代化建设提供支撑。