Analysis and Modeling of Diffuse Ultrasonic Signals for Structural Health Modeling

用于结构健康建模的漫射超声波信号分析和建模

• 批准号: 0401213
• 负责人: Jennifer Michaels
• 金额: $ 21万
• 依托单位: Georgia Tech Research Corporation
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2004
• 资助国家: 美国
• 起止时间: 2004-05-01 至 2008-04-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=0401213&HistoricalAwards=false
• 关键词: Analysis; Modeling; Diffuse; Ultrasonic; Signals

Analysis and Modeling of Diffuse Ultrasonic Signalsfor Structural Health MonitoringThe use of permanently mounted sensors to monitor the health of critical structures such as airplanes,bridges and buildings is quickly becoming a reality as sensors for measuring such physical quantities astemperature, moisture and strain become smaller and more robust. However, these devices are limitedto point, or local, measurements and thus do not truly interrogate the volumetric state of the structure.Sparse arrays of permanently mounted ultrasonic sensors, acting as both sources and receivers, can sendultrasonic energy throughout the entire structural volume and thus have the potential to detect criticalchanges. Before this goal can become a reality, revolutionary advances must be made in the analysisand processing of the received ultrasonic signals so that structural changes that may lead to catastrophicfailure can be reliably detected with an acceptably low false alarm rate. A significant complication isthat benign environmental effects such as changes in temperature and surface conditions can causelarger changes in the ultrasonic signals than actual flaws.Intellectual Merit: The research proposed here is a combination of experiments and waveformmodeling, and considers the development and verification of signal processing, classification and datafusion methods based upon quantitative changes in the ultrasonic signals. The testing mode consideredis that of diffuse ultrasonic waves whereby a point-like impulsive excitation is used to generate multi-modalelastic waves that fill the structure with sound. Ultrasonic diffuse wave theory models the rateof energy decay but does not predict the details of the complex time domain signals. The proposedresearch will combine time-dependent analysis of coherence with feature extraction, classificationmethods, signal modeling and simulation, and data fusion to tackle the challenging problem of detectingand characterizing damage. One inherent problem in using classification methods such as neuralnetworks for this application is the need to have a large set of signals from a wide variety of flaws; thisis not practical for structures outside the laboratory. A key aspect of the proposed research is todevelop methods for perturbing the baseline signal from the undamaged structure in order to emulate awide variety of structural and environmental changes.Four tasks are defined as follows:1. Signal Processing and Classification Methods. Development of quantitative differential methods todetermine if signal changes are due to structural or environmental effects, and to characterizestructural changes as to location, severity, type, etc.2. Modeling and Simulation of Diffuse Ultrasonic Signals. Modeling of diffuse ultrasonic signals, andsimulation of environmental and structural changes.3. Transducer Placement and Data Fusion. Analysis of optimum transducer organization andplacement, and fusing of data from multiple transducers.4. Design and Implementation of Experimental Measurements. Ultrasonic diffuse wavemeasurements using metallic, composite and cement-based structures.Broader Impact: This multidisciplinary research program will lead to effective use of ultrasonicsensors for continuously monitoring the health of critical structures, which will enable appropriateaction to take place prior to catastrophic failure. Furthermore, the methodologies developed will havebroad application to other disciplines such as sonar, radar and biomedical signal processing. A key partof the educational impact is the participation of undergraduates as well as graduate students, and aconcerted effort will be made to recruit women and underrepresented minorities. Also proposed is thedevelopment of a graduate course in ultrasonic wave propagation and signal processing that willcombine the fundamentals of acoustic and elastic wave propagation with signal processing methods asapplied to ultrasonics. This research program will also complement other efforts at Georgia Tech,effectively creating a critical mass of research in ultrasonics for structural health monitoring.

用于结构健康监测的弥散超声信号分析和建模随着用于测量温度、湿度和应变等物理量的传感器变得越来越小，越来越坚固，使用永久性安装的传感器来监测飞机、桥梁和建筑物等关键结构的健康状况正迅速成为现实。然而，这些设备仅限于点或局部测量，因此不能真正询问结构的体积状态。永久安装的超声波传感器的稀疏阵列，作为源和接收器，可以在整个结构体积中发送超声波能量，因此具有检测关键变化的潜力。在这一目标成为现实之前，必须在接收到的超声波信号的分析和处理方面取得革命性的进步，以便能够以可接受的低误报率可靠地检测到可能导致灾难性故障的结构变化。一个重要的并发症是，温和的环境影响，如温度和表面条件的变化，会导致超声波信号的变化比实际缺陷更大。智力优势：本文提出的研究是实验与波形建模相结合，并考虑基于超声信号定量变化的信号处理、分类和数据融合方法的开发和验证。所考虑的测试模式是漫射超声波，即使用点状脉冲激励产生多模态弹性波，使结构充满声音。超声漫射波理论建立了能量衰减率的模型，但不能预测复杂时域信号的细节。提出的研究将结合相干性的时间相关分析与特征提取、分类方法、信号建模和仿真以及数据融合，以解决检测和表征损伤的挑战性问题。在这种应用中使用神经网络等分类方法的一个固有问题是，需要从各种各样的缺陷中获得大量的信号；这对于实验室之外的结构是不实用的。提出的研究的一个关键方面是开发方法来干扰基线信号从未损坏的结构，以模拟各种各样的结构和环境变化。定义如下四个任务：1。信号处理与分类方法。发展定量差分方法，以确定信号变化是由于结构影响还是环境影响，并描述结构变化的位置、严重程度、类型等。漫射超声信号的建模与仿真。扩散超声信号的建模，以及环境和结构变化的模拟。传感器安置和数据融合。3 .传感器的最佳组织和位置分析，以及多传感器数据的融合。实验测量的设计与实现。使用金属、复合材料和水泥基结构的超声波漫射波测量。更广泛的影响：这个多学科研究项目将有效地使用超声波传感器来持续监测关键结构的健康状况，这将使在灾难性故障之前采取适当的行动。此外，所开发的方法将广泛应用于其他学科，如声纳，雷达和生物医学信号处理。教育影响的一个关键部分是本科生和研究生的参与，并将作出协调一致的努力招收妇女和代表性不足的少数民族。同时建议开设超声波传播和信号处理研究生课程，将声波和弹性波传播的基本原理与应用于超声波的信号处理方法结合起来。该研究项目还将补充佐治亚理工学院的其他工作，有效地为结构健康监测的超声研究创造一个临界质量。