Though automatic structural damage inspection methods based on mobile robots have been rapidly developed recently, they are not able to completely replaced conventional human based inspections due to limitations such as the false alarm of automatic damage detection techniques. In this study, a human centered inspection system with aid of an augmented reality framework is developed in order to improve convenience in inspecting and managing various types of structural damages such as spalling, exposed rebars and efflorescence. The developed system automatically detects and quantifies structural damages, and displays the inspection results in real-time through an augmented reality device. In addition, the previously detected damages are visualized with holographic markers and their information at the exact location. Therefore, an inspector can easily find where the previous damages were and whether the damages become severe or not. The performance of the developed system was validated through a field test and it was revealed that the system can save inspection time and improve convenience by accelerating essential tasks of the inspector such as damage detection, size measurement and finding locations of previous damages and determining whether the damages become severe or not.
{"title":"DEVELOPMENT OF STRUCTURAL DAMAGE INSPECTION AND MAINTENANCE SYSTEM BASED ON MIXED REALITY","authors":"Junyeon Chung, H. Sohn","doi":"10.12783/shm2021/36234","DOIUrl":"https://doi.org/10.12783/shm2021/36234","url":null,"abstract":"Though automatic structural damage inspection methods based on mobile robots have been rapidly developed recently, they are not able to completely replaced conventional human based inspections due to limitations such as the false alarm of automatic damage detection techniques. In this study, a human centered inspection system with aid of an augmented reality framework is developed in order to improve convenience in inspecting and managing various types of structural damages such as spalling, exposed rebars and efflorescence. The developed system automatically detects and quantifies structural damages, and displays the inspection results in real-time through an augmented reality device. In addition, the previously detected damages are visualized with holographic markers and their information at the exact location. Therefore, an inspector can easily find where the previous damages were and whether the damages become severe or not. The performance of the developed system was validated through a field test and it was revealed that the system can save inspection time and improve convenience by accelerating essential tasks of the inspector such as damage detection, size measurement and finding locations of previous damages and determining whether the damages become severe or not.","PeriodicalId":180083,"journal":{"name":"Proceedings of the 13th International Workshop on Structural Health Monitoring","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-03-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122275985","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
New offshore wind farms are often operating far from the shore and under challenging operating conditions, making manual on-site inspections expensive. Therefore, there is a growing need for remote condition monitoring and prognostics systems for such offshore wind farms. In this paper, we focus on corrosion prognosis since corrosion is a major failure mode of offshore wind turbine structures. In particular, we propose two algorithms for corrosion prognosis by employing Bayesian filtering techniques, one is based on linear degradation and another is based on a bi-modal corrosion model. Due to distinct characteristics of the two degradation models, different Bayesian filtering implementations are therefore required. Although the degradation model of the latter method more accurately reflects the ground truth, we find that the former prognosis method is computationally more efficient and likely more robust against various noise sources.
{"title":"CORROSION PROGNOSTICS FOR OFFSHORE WIND- TURBINE STRUCTURES USING BAYESIAN FILTERING WITH BI-MODAL AND LINEAR DEGRADATION MODELS","authors":"R. Brijder, Stijn Helsen, A. Ompusunggu","doi":"10.12783/shm2021/36288","DOIUrl":"https://doi.org/10.12783/shm2021/36288","url":null,"abstract":"New offshore wind farms are often operating far from the shore and under challenging operating conditions, making manual on-site inspections expensive. Therefore, there is a growing need for remote condition monitoring and prognostics systems for such offshore wind farms. In this paper, we focus on corrosion prognosis since corrosion is a major failure mode of offshore wind turbine structures. In particular, we propose two algorithms for corrosion prognosis by employing Bayesian filtering techniques, one is based on linear degradation and another is based on a bi-modal corrosion model. Due to distinct characteristics of the two degradation models, different Bayesian filtering implementations are therefore required. Although the degradation model of the latter method more accurately reflects the ground truth, we find that the former prognosis method is computationally more efficient and likely more robust against various noise sources.","PeriodicalId":180083,"journal":{"name":"Proceedings of the 13th International Workshop on Structural Health Monitoring","volume":"34 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-03-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132965317","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The coke drum, work under high temperature, coke in and out with hot and cold distress through the cycle of 500℃ to 50℃, is the core reactor of the delayed coking device. Typical failure modes of bulge deformation of tower, low cycle fatigue crack of weld, material aging, the tower body bending and inclining is easy to be caused under the comprehensive impact of complex load condition, temperature stress and mechanical stress of coke drums, which is a serious threat to safety production. The development and application of coke drum structural health monitoring system ensures the healthy operation of the equipment, and will effectively cope with the increasing long-term operation requirements of refining and chemical enterprises. Based on the stress monitoring data of the key parts of the coke drum, as well as the operating pressure and temperature data, this paper analyzes the characteristics of coke drum structure monitoring data and the causes of circumferential weld cracking of coke drum structure. The residual life of the monitored structure is evaluated by using the improved rain flow counting statistical method and miner fatigue cumulative damage theory. Finally, the contribution rate of different process sections of coke drum to its structural damage is analyzed. This study provides a reference for structural health monitoring and operation process optimization of coke drum.
{"title":"RESEARCH ON STRUCTURAL HEALTH DIAGNOSIS TECHNOLOGY OF COKE DRUM BASED ON MONITORING DATA","authors":"Fangxiong Tang, Dongming Yang, K. Ding, Li Chen","doi":"10.12783/shm2021/36347","DOIUrl":"https://doi.org/10.12783/shm2021/36347","url":null,"abstract":"The coke drum, work under high temperature, coke in and out with hot and cold distress through the cycle of 500℃ to 50℃, is the core reactor of the delayed coking device. Typical failure modes of bulge deformation of tower, low cycle fatigue crack of weld, material aging, the tower body bending and inclining is easy to be caused under the comprehensive impact of complex load condition, temperature stress and mechanical stress of coke drums, which is a serious threat to safety production. The development and application of coke drum structural health monitoring system ensures the healthy operation of the equipment, and will effectively cope with the increasing long-term operation requirements of refining and chemical enterprises. Based on the stress monitoring data of the key parts of the coke drum, as well as the operating pressure and temperature data, this paper analyzes the characteristics of coke drum structure monitoring data and the causes of circumferential weld cracking of coke drum structure. The residual life of the monitored structure is evaluated by using the improved rain flow counting statistical method and miner fatigue cumulative damage theory. Finally, the contribution rate of different process sections of coke drum to its structural damage is analyzed. This study provides a reference for structural health monitoring and operation process optimization of coke drum.","PeriodicalId":180083,"journal":{"name":"Proceedings of the 13th International Workshop on Structural Health Monitoring","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-03-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131103886","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Jitish Miglani, Wei Zhao, R. Kapania, Shardul S. Panwar, Rikin Gupta, A. Aris
The main objective of the presented work is to understand the aeroelastic characteristics of an inflatable kite. Experimental modal analyses of two test articles are conducted using system identification and signal processing. To understand the flutter behavior of inflatable wings, flutter test articles made from Nylon and Dyneema fabrics are tested first for their modal properties and then for flutter in a wind tunnel. Various experimental techniques are implemented for understanding the modal responses of these two test articles. Instruments such as a 3D-photogrammetry camera system and accelerometers are used to measure the dynamic and static responses of these test articles. Using MATLAB's System Identification Toolbox and Signal Processing Toolbox, the modal parameters are identified from measured responses, such as out-ofplane displacement and accelerations. The experimental and operational modal parameters are then used to estimate the modal responses. The Zimmerman- Weissenburger flutter margin parameter is used to predict the onset of the flutter modes from the identified modal parameters. The verified system identification technologies are leveraged to understand the aeroelastic dynamic instability of a tethered inflatable wing during a wind tunnel test.
{"title":"SYSTEM IDENTIFICATION FOR MODAL AND FLUTTER ANALYSIS OF AN INFLATABLE WING","authors":"Jitish Miglani, Wei Zhao, R. Kapania, Shardul S. Panwar, Rikin Gupta, A. Aris","doi":"10.12783/shm2021/36266","DOIUrl":"https://doi.org/10.12783/shm2021/36266","url":null,"abstract":"The main objective of the presented work is to understand the aeroelastic characteristics of an inflatable kite. Experimental modal analyses of two test articles are conducted using system identification and signal processing. To understand the flutter behavior of inflatable wings, flutter test articles made from Nylon and Dyneema fabrics are tested first for their modal properties and then for flutter in a wind tunnel. Various experimental techniques are implemented for understanding the modal responses of these two test articles. Instruments such as a 3D-photogrammetry camera system and accelerometers are used to measure the dynamic and static responses of these test articles. Using MATLAB's System Identification Toolbox and Signal Processing Toolbox, the modal parameters are identified from measured responses, such as out-ofplane displacement and accelerations. The experimental and operational modal parameters are then used to estimate the modal responses. The Zimmerman- Weissenburger flutter margin parameter is used to predict the onset of the flutter modes from the identified modal parameters. The verified system identification technologies are leveraged to understand the aeroelastic dynamic instability of a tethered inflatable wing during a wind tunnel test.","PeriodicalId":180083,"journal":{"name":"Proceedings of the 13th International Workshop on Structural Health Monitoring","volume":"3 1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-03-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115727278","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Yuguang Fu, Zixin Wang, A. Maghareh, S. Dyke, M. Jahanshahi, A. Shahriar
Due to their unpredictable nature, many impact events (e.g., overheight vehicles striking on low-clearance bridges) go unnoticed or get reported hours or days later. However, they can induce structural damage or even failure. Therefore, prompt impact detection and localization strategies are essential for early warning of impact events and rapid inspection of structures. Most existing strategies are developed for aircraft composites panels utilizing high rate synchronized measurement from densely deployed sensors. Limited efforts are made for other applications, such as infrastructure systems or extraterrestrial human habitats, which require large-scale measurement and scalable detection strategies. Particularly in harsh environments, structural impact localization must be robust to limited number of sensors and multi-source errors. In this study, an effective impact localization strategy is proposed to identify impact locations using limited number of vibration measurements. Convolutional neural networks are trained for each sensor node and are fused using Bayesian theory to improve the accuracy of impact localization. Special considerations are paid to address both measurement and modeling errors. The proposed strategy is illustrated using a 1D structure, and numerically validated for a 2D dome-shaped structure. The results demonstrate that the proposed method detects and localizes impact events accurately and robustly.
{"title":"SCALABLE IMPACT DETECTION AND LOCALIZATION USING DEEP LEARNING AND INFORMATION FUSION","authors":"Yuguang Fu, Zixin Wang, A. Maghareh, S. Dyke, M. Jahanshahi, A. Shahriar","doi":"10.12783/shm2021/36285","DOIUrl":"https://doi.org/10.12783/shm2021/36285","url":null,"abstract":"Due to their unpredictable nature, many impact events (e.g., overheight vehicles striking on low-clearance bridges) go unnoticed or get reported hours or days later. However, they can induce structural damage or even failure. Therefore, prompt impact detection and localization strategies are essential for early warning of impact events and rapid inspection of structures. Most existing strategies are developed for aircraft composites panels utilizing high rate synchronized measurement from densely deployed sensors. Limited efforts are made for other applications, such as infrastructure systems or extraterrestrial human habitats, which require large-scale measurement and scalable detection strategies. Particularly in harsh environments, structural impact localization must be robust to limited number of sensors and multi-source errors. In this study, an effective impact localization strategy is proposed to identify impact locations using limited number of vibration measurements. Convolutional neural networks are trained for each sensor node and are fused using Bayesian theory to improve the accuracy of impact localization. Special considerations are paid to address both measurement and modeling errors. The proposed strategy is illustrated using a 1D structure, and numerically validated for a 2D dome-shaped structure. The results demonstrate that the proposed method detects and localizes impact events accurately and robustly.","PeriodicalId":180083,"journal":{"name":"Proceedings of the 13th International Workshop on Structural Health Monitoring","volume":"438 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-03-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122148002","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Load rating of bridges is used to understand the working status and carrying capacity of bridge structures and components and is necessary to the safety of transportation. The current manual load rating procedure is, however, time-consuming. An intelligent and automatic load rating approach can be beneficial to supplement or eventually perhaps replace the current manual procedures. The innovation of this paper lies in developing an autonomous load rating framework by leveraging Digital Twin (DT) techniques. Full-scale laboratory testing of a bridge slab was conducted to verify the efficiency of the proposed framework. The ultimate moment capacity of the slab was obtained by carrying out four-point bending test. The testing procedure was monitored in real-time with multiple strain gauges. A real-scale finite element model of the slab was developed and calibrated with the testing results. The proposed DT framework of the bridge slabs was developed by integrating the numerical modeling and the strain monitoring. The proposed DT framework is intended for field application, and field results will be discussed.
{"title":"AN AUTONOMOUS BRIDGE LOAD RATING FRAMEWORK USING DIGITAL TWIN","authors":"Li Ai, M. Bayat, G. Comert, P. Ziehl","doi":"10.12783/shm2021/36329","DOIUrl":"https://doi.org/10.12783/shm2021/36329","url":null,"abstract":"Load rating of bridges is used to understand the working status and carrying capacity of bridge structures and components and is necessary to the safety of transportation. The current manual load rating procedure is, however, time-consuming. An intelligent and automatic load rating approach can be beneficial to supplement or eventually perhaps replace the current manual procedures. The innovation of this paper lies in developing an autonomous load rating framework by leveraging Digital Twin (DT) techniques. Full-scale laboratory testing of a bridge slab was conducted to verify the efficiency of the proposed framework. The ultimate moment capacity of the slab was obtained by carrying out four-point bending test. The testing procedure was monitored in real-time with multiple strain gauges. A real-scale finite element model of the slab was developed and calibrated with the testing results. The proposed DT framework of the bridge slabs was developed by integrating the numerical modeling and the strain monitoring. The proposed DT framework is intended for field application, and field results will be discussed.","PeriodicalId":180083,"journal":{"name":"Proceedings of the 13th International Workshop on Structural Health Monitoring","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-03-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130142560","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Monte Carlo sampling approaches are frequently used for probabilistic model updating of physics-based models under parametric uncertainty due to their high accuracy. The model updating framework produces a model that represents the real system more accurately than the prior knowledge or assumptions. This statistically updated model may prove useful if Structural Health Monitoring (SHM) techniques are to be applied. However, the updating of the models requires the use of a high number of samples, implying a high computational cost. Another additional disadvantage of these methods is that most of them require the calibration of a high number of parameters for their algorithm to become sampling efficient. Variational inference (VI) is an alternative approach for inference often used by the machine learning community. An optimization algorithm is employed to choose from a family of distributions the member that best approximates the posterior. In the method described in this paper the variational posterior that maximises the evidence lower bound (ELBO) is chosen. An approach based on VI is proposed and implemented on two different numerical examples to infer the uncertain parameters by postulating a variational posterior distribution given by a multivariate Gaussian approximation. It has been found that the number of samples required for the calculation of the posterior is reduced compared with Monte Carlo sampling approaches, however this occurs at the cost of some accuracy. The methodology will be helpful for the development of enhanced SHM strategies that require fast inference under a limited computational budget.
{"title":"STRUCTURAL MODEL UPDATING USING VARIATIONAL INFERENCE","authors":"F. Igea, M. Chatzis, A. Cicirello","doi":"10.12783/shm2021/36282","DOIUrl":"https://doi.org/10.12783/shm2021/36282","url":null,"abstract":"Monte Carlo sampling approaches are frequently used for probabilistic model updating of physics-based models under parametric uncertainty due to their high accuracy. The model updating framework produces a model that represents the real system more accurately than the prior knowledge or assumptions. This statistically updated model may prove useful if Structural Health Monitoring (SHM) techniques are to be applied. However, the updating of the models requires the use of a high number of samples, implying a high computational cost. Another additional disadvantage of these methods is that most of them require the calibration of a high number of parameters for their algorithm to become sampling efficient. Variational inference (VI) is an alternative approach for inference often used by the machine learning community. An optimization algorithm is employed to choose from a family of distributions the member that best approximates the posterior. In the method described in this paper the variational posterior that maximises the evidence lower bound (ELBO) is chosen. An approach based on VI is proposed and implemented on two different numerical examples to infer the uncertain parameters by postulating a variational posterior distribution given by a multivariate Gaussian approximation. It has been found that the number of samples required for the calculation of the posterior is reduced compared with Monte Carlo sampling approaches, however this occurs at the cost of some accuracy. The methodology will be helpful for the development of enhanced SHM strategies that require fast inference under a limited computational budget.","PeriodicalId":180083,"journal":{"name":"Proceedings of the 13th International Workshop on Structural Health Monitoring","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-03-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130649654","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Given the general condition of road infrastructure in Costa Rica, the proper prioritization of maintenance activities for bridges is essential for government institutions to effectively plan and assign resource investments. This work presents the main results of an extension project developed by the e-Bridge program of the Costa Rica Institute of Technology, with the objective of designing and applying a methodology for prioritizing maintenance activities for bridges, taking as a case study the actual bridges managed by a specific regional municipality. To this end, first, a given set of bridges were inspected and evaluated. Then, with this detailed inventory information, a set of key bridge performance indicators were defined including structural condition, environmental variables, and socio-economical categories. Consequently, a tailor-made methodology was proposed to prioritize different kinds of maintenance activities for the respective bridges using the above-mentioned indicators. The methodology was implemented using a business intelligence tool to manage all the information and solve prioritization queries. This tool and the major findings of the project were shared during the project with community actors and municipality collaborators through several workshops. The resulting methodology and developed tool effectively support decision-making regarding bridge maintenance activities for the target municipality and could be applied nation-wide.
{"title":"A BUSINESS INTELLIGENCE APPROACH TO PRIORITIZE BRIDGE MAINTENANCE ACTIVITIES","authors":"Giannina Ortiz, C. Garita","doi":"10.12783/shm2021/36245","DOIUrl":"https://doi.org/10.12783/shm2021/36245","url":null,"abstract":"Given the general condition of road infrastructure in Costa Rica, the proper prioritization of maintenance activities for bridges is essential for government institutions to effectively plan and assign resource investments. This work presents the main results of an extension project developed by the e-Bridge program of the Costa Rica Institute of Technology, with the objective of designing and applying a methodology for prioritizing maintenance activities for bridges, taking as a case study the actual bridges managed by a specific regional municipality. To this end, first, a given set of bridges were inspected and evaluated. Then, with this detailed inventory information, a set of key bridge performance indicators were defined including structural condition, environmental variables, and socio-economical categories. Consequently, a tailor-made methodology was proposed to prioritize different kinds of maintenance activities for the respective bridges using the above-mentioned indicators. The methodology was implemented using a business intelligence tool to manage all the information and solve prioritization queries. This tool and the major findings of the project were shared during the project with community actors and municipality collaborators through several workshops. The resulting methodology and developed tool effectively support decision-making regarding bridge maintenance activities for the target municipality and could be applied nation-wide.","PeriodicalId":180083,"journal":{"name":"Proceedings of the 13th International Workshop on Structural Health Monitoring","volume":"145 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-03-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131982329","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
High-rate, high-acceleration dynamic events produce especially limited and sparse data for two main reasons: high-acceleration loadings can destroy the test article, and the required laboratory equipment is typically expensive and complicated to operate. In many cases, these limitations prevent researchers from collecting additional data, driving the need for machine learning algorithms that utilize small datasets. Despite deep learning’s preference for thousands or millions of training examples, the dataset considered in this work contains only six independent examples. Finite element analysis software simulates the dynamic response of an electronic structure, supplementing this small dataset with additional training examples. A hybrid deep learning model first learns the dynamic response of the simulated structure and is then adapted to predict the actual electronic structure’s damage levels. This work shows that physics-enhanced transfer learning improves structural damage classification accuracy (𝑃 = 0.0879).
{"title":"PHYSICS-ENHANCED DAMAGE CLASSIFICATION OF SPARSE DATASETS USING TRANSFER LEARNING","authors":"M. Todisco, Z. Mao","doi":"10.12783/shm2021/36292","DOIUrl":"https://doi.org/10.12783/shm2021/36292","url":null,"abstract":"High-rate, high-acceleration dynamic events produce especially limited and sparse data for two main reasons: high-acceleration loadings can destroy the test article, and the required laboratory equipment is typically expensive and complicated to operate. In many cases, these limitations prevent researchers from collecting additional data, driving the need for machine learning algorithms that utilize small datasets. Despite deep learning’s preference for thousands or millions of training examples, the dataset considered in this work contains only six independent examples. Finite element analysis software simulates the dynamic response of an electronic structure, supplementing this small dataset with additional training examples. A hybrid deep learning model first learns the dynamic response of the simulated structure and is then adapted to predict the actual electronic structure’s damage levels. This work shows that physics-enhanced transfer learning improves structural damage classification accuracy (𝑃 = 0.0879).","PeriodicalId":180083,"journal":{"name":"Proceedings of the 13th International Workshop on Structural Health Monitoring","volume":"41 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-03-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130439669","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Among the many methods proposed in the literature to perform structural health monitoring (SHM) of thin-walled structures, two of them appear to be particularly promising and complementary. On the one hand, integrating Machine Learning techniques into this field seems a remarkable solution, since these methods have been shown to be effective in recognising usually hard-to-detect recurring patterns in the measured signals related to the presence of damages in structures, thus improving the diagnostic performances of SHM frameworks. In particular, in the past years, Deep Learning algorithms have gained much importance in this field due to their capability of processing high-dimensional inputs (such as images), thus making it possible to automatically identify onsetting structural damages. On the other hand, ultrasonic guided wave-based approaches are commonly adopted to assess the structural integrity of plate-like structures and pipelines. These approaches, coupled with tomographic algorithms, typically allow performing damage detection and localisation with satisfactory results. However, such reconstruction algorithms are significantly sensors layout-dependent and, as such, they come with some still unsolved issues, leading, for example, to artifacts creation and unsatisfactory tomographic damage localisation performances in case of unevenly distributed network of sensors or when few sensors are installed on the structure. In this work, convolutional neural networks (CNNs) and ultrasonic guided waves are combined into a unique framework, which leverages on the advantages of the two methods to perform damage detection and localisation in platelike structures. Guided waves are excited and sensed by a network of sensors permanently installed on the structure. The information acquired is then converted into grayscale image as is, without performing any prior feature extraction procedure, which is further analysed by a set of CNNs. First, a classifier is employed to perform damage detection. In case damage is identified, the grayscale image is then analysed by two regression CNNs to localise the damage. The framework is tested using experimentally validated numerical simulations of guided waves propagating in a metallic plate available in the literature.
{"title":"CONVOLUTIONAL NEURAL NETWORKS FOR ULTRASONIC GUIDED WAVE-BASED STRUCTURAL DAMAGE DETECTION AND LOCALISATION","authors":"L. Lomazzi, M. Giglio, F. Cadini","doi":"10.12783/shm2021/36301","DOIUrl":"https://doi.org/10.12783/shm2021/36301","url":null,"abstract":"Among the many methods proposed in the literature to perform structural health monitoring (SHM) of thin-walled structures, two of them appear to be particularly promising and complementary. On the one hand, integrating Machine Learning techniques into this field seems a remarkable solution, since these methods have been shown to be effective in recognising usually hard-to-detect recurring patterns in the measured signals related to the presence of damages in structures, thus improving the diagnostic performances of SHM frameworks. In particular, in the past years, Deep Learning algorithms have gained much importance in this field due to their capability of processing high-dimensional inputs (such as images), thus making it possible to automatically identify onsetting structural damages. On the other hand, ultrasonic guided wave-based approaches are commonly adopted to assess the structural integrity of plate-like structures and pipelines. These approaches, coupled with tomographic algorithms, typically allow performing damage detection and localisation with satisfactory results. However, such reconstruction algorithms are significantly sensors layout-dependent and, as such, they come with some still unsolved issues, leading, for example, to artifacts creation and unsatisfactory tomographic damage localisation performances in case of unevenly distributed network of sensors or when few sensors are installed on the structure. In this work, convolutional neural networks (CNNs) and ultrasonic guided waves are combined into a unique framework, which leverages on the advantages of the two methods to perform damage detection and localisation in platelike structures. Guided waves are excited and sensed by a network of sensors permanently installed on the structure. The information acquired is then converted into grayscale image as is, without performing any prior feature extraction procedure, which is further analysed by a set of CNNs. First, a classifier is employed to perform damage detection. In case damage is identified, the grayscale image is then analysed by two regression CNNs to localise the damage. The framework is tested using experimentally validated numerical simulations of guided waves propagating in a metallic plate available in the literature.","PeriodicalId":180083,"journal":{"name":"Proceedings of the 13th International Workshop on Structural Health Monitoring","volume":"154 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-03-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126892052","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：