Long-term health monitoring of concrete and steel bridges under large and missing data by unsupervised meta learning

Engineering Structures, Journal Year: 2023, Volume and Issue: 279, P. 115616 - 115616

Published: Jan. 17, 2023

Long-term monitoring brings an important benefit for health of civil structures due to covering all possible unpredictable variations in measured vibration data and providing relatively adequate training samples unsupervised learning algorithms. Despite such merits, this process may encounter large with missing values also yield erroneous results caused by severe environmental changes, particularly those emerge as sharp increases modal frequencies during freezing weather. To address these challenges, article proposes a novel meta-learning method that entails four steps initial analysis, segmentation, subspace searching approach called nearest cluster selection, anomaly detection. The first step intends initially analyze data/features cleaning samples. Next, the second exploits spectral clustering divide clean into some segments. In third step, proposed selection is utilized measure dissimilarities between segments distance metric select minimum representative main segment. Finally, locally robust Mahalanobis-squared applied merging concepts statistics local online key innovations research contain developing new strategy alongside proposing idea selection. full-scale concrete steel bridges are used verify method. Results demonstrate succeeds mitigating effects accurately detecting damage.

Language: Английский

---

Review on smartphone sensing technology for structural health monitoring

Measurement, Journal Year: 2023, Volume and Issue: 223, P. 113716 - 113716

Published: Oct. 25, 2023

Sensing is a critical and inevitable sector of structural health monitoring (SHM). Recently, smartphone sensing technology has become an emerging, affordable, effective system for SHM other engineering fields. This because modern equipped with various built-in sensors technologies, especially triaxial accelerometer, gyroscope, global positioning system, high-resolution cameras, wireless data communications under the internet-of-things paradigm, which are suitable vibration- vision-based applications. article presents state-of-the-art review on recent research progress smartphone-based SHM. Although there some short reviews this topic, major contribution to exclusively present comprehensive survey practices civil structures from perspectives measurement techniques, third-party apps developed in Android iOS, application domains. Findings provide thorough understanding main ideas studies technology.

Language: Английский

---

A novel double-hybrid learning method for modal frequency-based damage assessment of bridge structures under different environmental variation patterns

Mechanical Systems and Signal Processing, Journal Year: 2023, Volume and Issue: 201, P. 110676 - 110676

Published: Aug. 10, 2023

Monitoring of modal frequencies under an unsupervised learning framework is a practical strategy for damage assessment civil structures, especially bridges. However, the key challenge related to high sensitivity environmental and/or operational changes that may lead economic and safety losses. The other pertains different variation patterns in due differences structural types, materials, applications, measurement periods terms short long monitoring programs, geographical locations weather conditions, influences single or multiple factors, which cause barriers employing state-of-the-art approaches. To cope with these issues, this paper proposes novel double-hybrid technique manner. It contains two stages data partitioning anomaly detection, both comprise hybrid algorithms. For first stage, improved clustering method based on coupling shared nearest neighbor searching density peaks proposed prepare local information detection focus mitigating effects. second innovative non-parametric detector outlier factor. In stages, number neighbors hyperparameter automatically determined by leveraging self-adaptive algorithm. Modal full-scale bridges are utilized validate several comparisons. Results indicate able successfully eliminate variations correctly detect damage.

Language: Английский

---

Abnormal data detection for structural health monitoring: State-of-the-art review

Developments in the Built Environment, Journal Year: 2024, Volume and Issue: 17, P. 100337 - 100337

Published: Jan. 19, 2024

Structural health monitoring (SHM) is widely used to monitor and assess the condition performance of engineering structures such as, buildings, bridges, dams, tunnels. Owing sensor defects, data acquisition errors, environmental interference, abnormal are often collected stored in systems. The this study essentially different from so-called "abnormal state data," which result structural physical damage or degradation. Abnormal totally related external interference rather than changes inherent features. However, can significantly affect assessment structures. It imperative detect remove measurements avoid misjudging SHM. This paper summarizes detection SHM field discusses relevant challenges. Moreover, background knowledge regarding introduced. methods then classified into statistical probability methods, predictive models, computer vision methods. advantages, disadvantages, scope each method investigated. An example detecting for a cable-stayed bridge In addition, issues existing studies summarized, future interests discussed.

Language: Английский

---

Short-term damage alarming with limited vibration data in bridge structures: A fully non-parametric machine learning technique

Measurement, Journal Year: 2024, Volume and Issue: 235, P. 114935 - 114935

Published: May 21, 2024

Machine learning-assisted vibration monitoring is an intelligent, automated, and popular strategy for evaluating civil structures damage alarming. However, implementing this under a short-term program may encounter challenges such as limited data, profound environmental operational variations, the limitations of state-of-the-art solutions these conditions. The main purpose paper to propose novel machine learning technique in terms unsupervised alarming with data. crux lies two fully non-parametric parts data partitioning anomaly detection. Initially, clustering approach procedure presented divide into clusters. Subsequently, new density-based detector developed prepare indicators Limited eigenfrequencies full-scale bridge are used validate proposed solution. Results can substantiate its effectiveness practicability programs.

Language: Английский

---

On continuous health monitoring of bridges under serious environmental variability by an innovative multi-task unsupervised learning method

Structure and Infrastructure Engineering, Journal Year: 2023, Volume and Issue: 20(12), P. 1975 - 1993

Published: Jan. 16, 2023

Design of an automated and continuous framework is paramount importance to structural health monitoring (SHM). This study proposes innovative multi-task unsupervised learning method for early assessment damage in large-scale bridge structures under long-term monitoring. entails three main tasks data cleaning, partitioning, anomaly detection. The first task includes discarding missing providing outlier-free samples by developing approach based on the well-known DBSCAN algorithm. Accordingly, this enforces generate two clusters, one which contains other comprises outlier data. In second task, are fed into spectral clustering partition them local clusters. Subsequently, a cluster with maximum cumulative density selected as optimal whose features extracted representative Finally, empirical measures theory used compute indices SHM. Long-term modal frequencies full-scale bridges incorporated verify proposed alongside comparative analyses. Results prove that can effectively detect discriminative scores mitigating negative influences severe environmental variability.

Language: Английский

---

A locally unsupervised hybrid learning method for removing environmental effects under different measurement periods

Measurement, Journal Year: 2023, Volume and Issue: 208, P. 112465 - 112465

Published: Jan. 11, 2023

Language: Английский

---

Unsupervised data normalization for continuous dynamic monitoring by an innovative hybrid feature weighting-selection algorithm and natural nearest neighbor searching

Structural Health Monitoring, Journal Year: 2023, Volume and Issue: 22(6), P. 4005 - 4026

Published: April 1, 2023

Continuous dynamic monitoring brings an important opportunity to evaluate the health and integrity of civil structures in a long-term manner. However, high dimensionality sparsity data caused by negative influences environmental and/or operational variability are major challenges this process. To address these issues, article proposes innovative unsupervised normalization method based on novel hybrid feature weighting-selection algorithm idea natural nearest neighbor (NN) searching emanated from theory mutual friendships human societies. The proposed is combination global weighting with new measure local selection. For algorithm, leverages NN that seeks find adequate NNs automatically. main objective remove effects provide normalized weighted features for reliable continuous monitoring. Using such features, anomaly detector Mahalanobis-squared distance developed assess detect structural damage. key innovations paper contain proposing fully nonparametric learning technique two parts detection developing removing variations. Long-term (modal frequencies) three-span box-girder concrete bridge (Z24 Bridge) long-span arch (Infante Dom Henrique considered verify several comparisons. Results indicate successful mitigating notifying accurate states.

Language: Английский

---

Bridge monitoring: Application of the extreme function theory for damage detection on the I-40 case study

Engineering Structures, Journal Year: 2023, Volume and Issue: 279, P. 115573 - 115573

Published: Jan. 9, 2023

Language: Английский

---

Investigation of Temperature Effects into Long-Span Bridges via Hybrid Sensing and Supervised Regression Models

Remote Sensing, Journal Year: 2023, Volume and Issue: 15(14), P. 3503 - 3503

Published: July 12, 2023

Temperature is an important environmental factor for long-span bridges because it induces thermal loads on structural components that cause considerable displacements, stresses, and damage. Hence, critical to acquire up-to-date information the status, sustainability, serviceability of under daily seasonal temperature fluctuations. This paper intends investigate effects variability displacements obtained from remote sensing represent their relationship using supervised regression models. In contrast other studies in this field, one contributions leverage hybrid as a combination contact non-contact sensors measuring data responses. Apart temperature, unmeasured operational conditions may affect separately or simultaneously. For issue, incorporates correlation analysis between measured predictor (temperature) response (displacement) linear measure, Pearson coefficient, well nonlinear measures, namely Spearman Kendall coefficients maximal criterion, determine whether dominant Finally, three techniques based model, Gaussian process regression, support vector are considered model conduct prediction process. limited displacement related used demonstrate results research. The aim research assess realize contact-based installed bridge structure and/or factors sufficient if necessary consider further investigations.

Language: Английский

---