Article
Engineering, Mechanical
Jun Zhan, Chengkun Wu, Xiandong Ma, Canqun Yang, Qiucheng Miao, Shilin Wang
Summary: This paper proposes an unsupervised time-series anomaly detection approach that combines deep learning with multi-parameter relative variability detection, which can effectively detect anomalies in wind turbine nacelles.
MECHANICAL SYSTEMS AND SIGNAL PROCESSING
(2022)
Article
Green & Sustainable Science & Technology
Ziqi Wang, Changliang Liu, Feng Yan
Summary: This paper proposes a SCADA data-driven method for condition monitoring of wind turbines using incremental learning and multivariate state estimation technique. The method updates the monitoring model in real-time and improves computation efficiency through sample selection and dynamic downsampling. Experimental results show that the method maintains high accuracy and low false alarm rate in long-term operation, and detects potential faults in advance.
Article
Green & Sustainable Science & Technology
Xiaohang Jin, Zhuangwei Xu, Wei Qiao
Summary: This article proposes an ensemble approach to detect anomalies and diagnose faults in wind turbines, based on modeling and analyzing historical SCADA data from healthy wind turbines. The method can detect anomalies and diagnose faults before wind turbines have to be shut down for maintenance.
IEEE TRANSACTIONS ON SUSTAINABLE ENERGY
(2021)
Article
Engineering, Multidisciplinary
Ling Xiang, Penghe Wang, Xin Yang, Aijun Hu, Hao Su
Summary: A new method for fault detection of wind turbines is proposed in this paper, which combines CNN and LSTM networks trained on SCADA data to enhance accuracy and predictive effectiveness.
Article
Computer Science, Information Systems
Lipeng Zhu, Yue Song
Summary: This article develops an automatic data label calibration method to improve the accuracy of WT fault information using time series data analysis. By analyzing temporal similarities and the diversity of multiple variables, the proposed approach reliably calibrates most of the data labels and determines the remaining labels using a new method.
IEEE INTERNET OF THINGS JOURNAL
(2022)
Article
Engineering, Electrical & Electronic
Jiarui Liu, Xinli Li, Chaojie Li, Guotian Yang, Yaqi Li, Jing Qiu, Zhao Yang Dong
Summary: The article proposes a generalized Siamese NBM scheme that explores the characteristics between anomalies and normal behavior. By considering fault samples and designing a parameter-shared backbone and auxiliary regularization terms, the proposed scheme improves the reliability and performance of anomaly detection. The use of a density-based clustering algorithm and label correction further enhances the trustworthiness of the scheme.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2023)
Article
Engineering, Electrical & Electronic
Siyu Tao, Qingshan Xu, Andres Feijoo, Gang Zheng
Summary: A bi-level multi-objective optimization framework is proposed in this study to design the configuration of wind turbines and the topology of the electrical collector system in an offshore wind farm for better performance. The outer-layer model optimizes the daily profit rate, daily average capacity factor, and power quality, while the inner layer models determine the electrical system topology and generation schedule of other generators.
IEEE TRANSACTIONS ON SMART GRID
(2021)
Article
Automation & Control Systems
Xingchen Liu, Juan Du, Zhi-Sheng Ye
Summary: This article develops a novel condition monitoring and fault isolation system for wind turbines based on SCADA data. The article addresses challenges such as low sampling rate, time-varying working conditions, and lack of historical fault data. The system uses preprocessing and a global monitoring statistic to monitor the health status of the wind turbine and isolate faults without expert knowledge or historical data.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2022)
Article
Energy & Fuels
Yusi Shih
Summary: This paper introduces the reliance of modern offshore wind farms on SCADA systems and the current interest in openness of working environments. The author proposes solutions utilizing the IEC 60870-5-104 protocol for wind farms' data storage and traffic control. The paper demonstrates the effectiveness of these solutions and aims to enhance the openness of SCADA working environments.
Article
Engineering, Marine
Mingqiang Xu, Francis T. K. Au, Shuqing Wang, Zhenshuang Wang, Qian Peng, Huiyuan Tian
Summary: This paper investigates the dynamic response of a monopile offshore wind turbine under earthquake excitations and presents some unexpected findings, such as the earthquake allowing the identification of high-frequency bending and twisting modes and a possible increase in the first frequency of the wind turbine.
Article
Chemistry, Multidisciplinary
Joachim Verhelst, Inge Coudron, Agusmian Partogi Ompusunggu
Summary: This paper describes the development of a software tool for offshore wind turbine corrosion monitoring and highlights the importance of its tight integration with decision support tools.
APPLIED SCIENCES-BASEL
(2022)
Article
Engineering, Multidisciplinary
Jiarui Liu, Guotian Yang, Xinli Li, Shumin Hao, Yingming Guan, Yaqi Li
Summary: This paper introduces a novel deep generative method based on the convolutional neural network (CNN)-conditional variational auto-encoder (CVAE) for fault detection of wind turbines. The method combines the feature extraction ability of CVAE and CNN to improve the performance of fault detection through learning probability distribution models and conducting time-series feature extraction and reconstruction.
MEASUREMENT SCIENCE AND TECHNOLOGY
(2023)
Article
Engineering, Electrical & Electronic
Ping Wu, Yixuan Wang, Xujie Zhang, Jinfeng Gao, Lin Wang, Yichao Liu
Summary: This article proposes a Mogrifier long short-term memory autoencoder (MLSTM-AE) method to monitor blade breakage in wind turbines. The method calculates the Pearson correlation coefficient for variable selection and uses MLSTM layers to extract spatial-temporal information. By applying kernel density estimation, boundaries for blade breakage alerts are generated based on reconstruction errors, enabling effective monitoring of system dynamics.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2023)
Article
Engineering, Electrical & Electronic
Kai Zhang, Baoping Tang, Lei Deng, Xiaoxia Yu
Summary: The study presents a fault detection frame based on subspace reconstruction-based robust kernel principal component analysis (SR-RKPCA) model for wind turbines SCADA data. By utilizing RKPCA method, permutation entropy, and combined index, the stability, accuracy, and non-linear feature extraction capability of the wind turbine fault detection model are enhanced.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2021)
Article
Green & Sustainable Science & Technology
Christopher Jung, Dirk Schindler
Summary: The goal of this study is to quantify the meteorological, geographical, and technical properties of the current global offshore wind turbine fleet. The factors studied show high regional variability, with European wind turbine sites providing higher wind resources compared to Asian sites, where turbines operate in shallower water closer to the shores. These findings suggest that wind potential and siting criteria for wind turbines differ depending on the country.
RENEWABLE & SUSTAINABLE ENERGY REVIEWS
(2023)
Article
Engineering, Multidisciplinary
Jack Poole, Paul Gardner, Nikolaos Dervilis, Lawrence Bull, Keith Worden
Summary: This article discusses the limitation of labelled data in the practical application of structural health monitoring, and introduces transfer learning methods, specifically domain adaptation. By using statistic alignment, the performance degradation issue under class imbalance in traditional methods is addressed, and the effectiveness of statistic alignment is demonstrated in numerical and real-world scenarios.
STRUCTURAL HEALTH MONITORING-AN INTERNATIONAL JOURNAL
(2023)
Article
Chemistry, Analytical
Marcus Haywood-Alexander, Nikolaos Dervilis, Keith Worden, Robin S. Mills, Purim Ladpli, Timothy J. Rogers
Summary: Ultrasonic guided waves provide a convenient and practical approach to structural health monitoring and non-destructive evaluation, with dispersion curves describing the relationship between frequency and propagation characteristics. Accurate dispersion curve information is valuable in many guided wave-based strategies, and can be determined through experimental observations and a system identification procedure. This study uses a scanning-laser Doppler vibrometer to record Lamb wave propagation in a composite plate and performs a Bayesian analysis using the Markov-Chain Monte Carlo method to determine dispersion curve information and infer confidence in the predicted parameters. The probabilistic approach is shown to have advantages over traditional estimation methods.
Article
Chemistry, Physical
Frank H. G. Stolze, Keith Worden, Graeme Manson, Wieslaw J. Staszewski
Summary: The use of a diffused Lamb wave field is a challenging yet effective method for fatigue-crack detection in multi-riveted strap-joint aircraft panels. The panel is equipped with low-profile surface-bonded piezoceramic transducers, which extract information on fatigue damage through various amplitude characteristics of Lamb waves. Statistical outlier analysis is performed to detect damage, and simplified wave scattering modeling is used to explain complex response features. The results demonstrate the potential and limitations of this method for reliable fatigue-crack detection in complex aircraft components.
Article
Engineering, Mechanical
R. Nayek, A. B. Abdessalem, N. Dervilis, E. J. Cross, K. Worden
Summary: This paper addresses the problem of identifying single degree-of-freedom (SDOF) nonlinear mechanical oscillators with piecewise-linear (PWL) restoring forces. A Bayesian framework is used to formulate the identification task as concurrent model selection and parameter estimation, and a likelihood-free Approximate Bayesian Computation (ABC) scheme is followed. The proposed approach has been demonstrated to effectively select models and identify parameters of PWL systems using numerical examples and an experimental study.
MECHANICAL SYSTEMS AND SIGNAL PROCESSING
(2023)
Article
Engineering, Multidisciplinary
Chandula T. Wickramarachchi, Eoghan Maguire, Elizabeth J. Cross, Keith Worden
Summary: Population-based structural health monitoring expands the scope of structural health monitoring from a single structure to a group of structures, allowing inferences and knowledge transfer within and between populations. This paper focuses on assessing the similarity of structures at the beginning of the analysis chain using distance metrics, to quickly identify abnormalities, group similar structures, and inform further decisions.
STRUCTURAL HEALTH MONITORING-AN INTERNATIONAL JOURNAL
(2023)
Proceedings Paper
Engineering, Mechanical
Max D. Champneys, Gerorge Tsialiamanis, Timothy J. Rogers, Nikolaos Dervilis, Keith Worden
Summary: Linear modal analysis provides a robust framework for analyzing structural dynamic systems. However, when nonlinearities are present, it is necessary to develop a nonlinear variant of modal analysis. This paper compares three approaches for constructing nonlinear normal modes (NNM) and applies nonlinear system identification techniques to study a simulated system with nonlinearities. The results demonstrate that the generated NNMs can effectively decompose the system and maintain excellent reconstruction at lower excitation levels.
NONLINEAR STRUCTURES & SYSTEMS, VOL 1
(2023)
Proceedings Paper
Engineering, Mechanical
Rajdip Nayek, Mohamed Anis Ben Abdessalem, Nikolaos Dervilis, Elizabeth J. Cross, Keith Worden
Summary: This paper addresses the problem of simultaneous model selection and parameter estimation for dynamical systems with piecewise-linear stiffnesses. It proposes a likelihood-free Approximate Bayesian Computation (ABC) scheme with nested sampling to simplify the jump between model spaces and flexible identify the correct model and parameters for PWL-stiffness systems.
NONLINEAR STRUCTURES & SYSTEMS, VOL 1
(2023)
Proceedings Paper
Engineering, Mechanical
Thomas Simpson, George Tsialiamanis, Nikolaos Dervilis, Keith Worden, Eleni Chatzi
Summary: Linear modal analysis provides a complete framework for the dynamic analysis of simplified engineering systems, while nonlinear modal analysis maintains some key features of modal analysis. Nonlinear normal modes serve as a basis for constructing reduced-order models.
NONLINEAR STRUCTURES & SYSTEMS, VOL 1
(2023)
Proceedings Paper
Engineering, Mechanical
Keith Worden, David J. Wagg, Malcolm Scott
Summary: This chapter discusses the challenges of validating models of nonlinear bifurcating systems and proposes approaches to overcome them. Validation is crucial for ensuring the accuracy of a model in representing the target structure or system. However, validating nonlinear models becomes more complicated due to the presence of bifurcations. If a model fails to capture the bifurcation points, its predictions can be highly inaccurate even if it closely resembles the real system. The chapter presents a case study on a three-storey shear building structure with nonlinearity and demonstrates the importance of considering modeling uncertainties in the validation process to improve discrimination capability.
MODEL VALIDATION AND UNCERTAINTY QUANTIFICATION, VOL 3
(2023)
Proceedings Paper
Engineering, Civil
G. Delo, A. Bunce, E. J. Cross, J. Gosliga, D. Hester, C. Surace, K. Worden, D. S. Brennan
Summary: Population-based Structural Health Monitoring (PBSHM) is a method for addressing the issue of data scarcity in traditional SHM by assessing the similarity of structures. Knowledge transfer only occurs between similar structures. This paper demonstrates the application of PBSHM to complex real structures.
EUROPEAN WORKSHOP ON STRUCTURAL HEALTH MONITORING (EWSHM 2022), VOL 2
(2023)
Proceedings Paper
Engineering, Civil
Tristan Gowdridge, Elizabeth J. Cross, Nikolaos Dervilis, Keith Worden
Summary: This paper investigates the topological changes of the Z24 Bridge's natural frequencies before and after cointegration, with a focus on the nonlinear effects of temperature on the second natural frequency. By using cointegration to normalize the data and remove environmental and operational variations, the temperature effects on the natural frequency data are clearly observed. The paper employs topological data analysis to analyze the raw and cointegrated time series, constructing simplicial complexes and calculating topological properties such as persistent homology to determine the overall topological structure of the time series.
EUROPEAN WORKSHOP ON STRUCTURAL HEALTH MONITORING (EWSHM 2022), VOL 2
(2023)
Proceedings Paper
Engineering, Civil
Chandula T. Wickramarachchi, Julian Gosliga, Elizabeth J. Cross, Keith Worden
Summary: Population-based structural health monitoring (PBSHM) extends the concept of structural health monitoring from single structures to a group of structures. It assesses the similarity of structures within the population in order to form communities and networks. Graphical representations of structures and kernel-based methods are used to evaluate similarities within the population.
EUROPEAN WORKSHOP ON STRUCTURAL HEALTH MONITORING (EWSHM 2022), VOL 2
(2023)
Proceedings Paper
Engineering, Civil
Marcus Haywood-Alexander, Nikolaos Dervilis, Keith Worden, Timothy J. Rogers
Summary: Guided waves are gaining interest in SHM due to their distinct advantages. Accurate dispersion curves are invaluable for guided-wave-based localization strategies, and the wave speed in complex materials is dependent on the propagation angle. This study uses a scanning laser Doppler vibrometer and a novel Legendre polynomial expansion approach to extract dispersion curve points and determine material properties governing Lamb wave propagation at various angles in carbon-fiber composite plates.
EUROPEAN WORKSHOP ON STRUCTURAL HEALTH MONITORING (EWSHM 2022), VOL 3
(2023)
Proceedings Paper
Engineering, Civil
Weijiang Lin, Keith Worden, Elizabeth Cross
Summary: This paper proposes a data-based model for accurately predicting the response of turbines in wind farms to environmental changes, with the aim of reducing operation and maintenance costs. Through a simulated case study, the model demonstrates good predictive accuracy and potential as a low-cost wake field predictor.
EUROPEAN WORKSHOP ON STRUCTURAL HEALTH MONITORING (EWSHM 2022), VOL 3
(2023)
Proceedings Paper
Engineering, Civil
Jack Poole, Paul Gardner, Nikolaos Dervilis, Lawrence Bull, Keith Worden
Summary: This paper addresses the limitation of unavailability of labelled data in structural health monitoring through transfer learning in the form of domain adaptation. It proposes a novel statistic alignment method and instance-weighting strategy to tackle the partial domain adaptation problem.
EUROPEAN WORKSHOP ON STRUCTURAL HEALTH MONITORING (EWSHM 2022), VOL 3
(2023)
