A distance-aware approach for reliable out-of-distribution detection of wind turbine gearbox fault diagnosis DOI Creative Commons

Junli Zhou,

Yao Zhao

Frontiers in Energy Research, Journal Year: 2024, Volume and Issue: 12

Published: Nov. 28, 2024

Fault diagnosis of wind turbine gearbox is essential to ensure operational efficiency and prevent costly downtime. However, conventional deep learning models often struggle with domain shift, where the distribution testing data differs from that training data. This issue more pronounced out-of-distribution inputs—data outside conditions model was trained on. These challenges can lead unreliable diagnostic results potentially hazardous situations. To address this, we introduce Spectral Normalization Gaussian Process methods into Res2Net framework enhance its ability detect improve model’s assess distance between test handle due both epistemic aleatory uncertainty. The experiment collected raw vibration signals under varied conditions. Unknown faults simulated uncertainty, while noisy samples resulted in were converted images using Gramian Angular Difference Field transformation. resulting then fed model, enhanced Process. outputs include classification corresponding uncertainty values based on awareness. With quantified values, reflect trustworthiness results. By comparing these predefined thresholds, it possible distinguish whether are or not. Experiments have proven superiority Distance-Aware detection fault diagnosis.

Language: Английский

Multilevel feature encoder for transfer learning-based fault detection on acoustic signal DOI
Dezheng Wang, Congyan Chen

Information Fusion, Journal Year: 2025, Volume and Issue: unknown, P. 103128 - 103128

Published: March 1, 2025

Language: Английский

Citations

1
Digital twin-inspired methods for rotating machinery intelligent fault diagnosis and remain useful life prediction: A state-of-the-art review and future challenges DOI
Kun Yu, Caizi Fan, Yongchao Zhang

et al.

Mechanical Systems and Signal Processing, Journal Year: 2025, Volume and Issue: 232, P. 112770 - 112770

Published: April 21, 2025

Language: Английский

Citations

1
Revolutionizing wind turbine fault diagnosis on supervisory control and data acquisition system with transparent artificial intelligence DOI
Muhammad Irfan, Sana Yasin,

Umar Draz

et al.

International Journal of Green Energy, Journal Year: 2025, Volume and Issue: unknown, P. 1 - 17

Published: Jan. 7, 2025

This research proposes a new method for wind turbine fault detection using hybrid deep neural networks, along with explainable Artificial Intelligence (XAI) methods. unique combination delivers more accurate and interpretable model to enable improved maintenance strategies efficiency in operations. The novelty of this work is that it an intact methodology utilizing AI turbines overcome the known weaknesses existing methods terms transparency interpretability. aims create elaborate system capable accurately predicting faults providing engineers transparent comprehendible explanations regarding decisions undertaken, which will further contribute learning. observations were made during numerous simulations tests, proposed XAI-driven indicated significant increase 99% accuracy rate maintained level effectiveness upkeep. approach expected reduce frequency operation disruptions put as standard reliability industry, nests potential novel industrial asset would significantly redefine rules field.

Language: Английский

Citations

0
MCRCNet: A Bearing Fault Diagnosis Method for Unknown Faults Based on Transfer Learning DOI Creative Commons
Guangyuan Xu, Ruifeng Guo, Zhenyu Yin

et al.

Applied Sciences, Journal Year: 2025, Volume and Issue: 15(2), P. 921 - 921

Published: Jan. 18, 2025

Bearing fault diagnosis in actual working conditions often faces the problem that unknown type faults cannot be identified, which seriously restricts practical application of technology. To solve this problem, paper proposes a bearing method based on transfer learning. Firstly, designs feature extraction network, Multi-scale Convolution-Convolutional Reconstruction Network (MCRCNet), incorporates multi-scale module to extract features at multiple scales, thereby enhancing ability key information. Secondly, an improved convolutional reconstruction AcConv (Adaptive Convolution reconstruction), highlights information and reduces redundant by reconstructing map. Furthermore, also modifies loss function improve performance case data imbalance, introduces Wasserstein distance optimize adversarial training process. The proposed is experimentally verified Case Western Reserve University, Jiangnan laboratory datasets. experimental results show has good most tasks generalization ability, provides feasible solution for research diagnosis.

Language: Английский

Citations

0
Fault Warning Study of Gearbox Based on SOM-ASTGCN-BiLSTM and Mutual Diagnosis of Same Clustered Wind Turbines DOI
Bo Gu, Hongtao Zhang, Shuai Yue

et al.

Renewable Energy, Journal Year: 2025, Volume and Issue: unknown, P. 123442 - 123442

Published: May 1, 2025

Language: Английский

Citations

0
A fault diagnosis method for offshore wind turbine bearing based on adaptive deep echo state network and bidirectional long short term memory network in noisy environment DOI

Yuanhao Du,

Xiuli Geng, Qingchao Zhou

et al.

Ocean Engineering, Journal Year: 2024, Volume and Issue: 312, P. 119101 - 119101

Published: Sept. 2, 2024

Language: Английский

Citations

3
Wind Turbine Blade Fault Detection Method Based on TROA-SVM DOI Creative Commons

Zhuo Lei,

Haijun Lin, Xudong Tang

et al.

Sensors, Journal Year: 2025, Volume and Issue: 25(3), P. 720 - 720

Published: Jan. 24, 2025

Wind turbines are predominantly situated in remote, high-altitude regions, where they face a myriad of harsh environmental conditions. Factors such as high humidity, strong gusts, lightning strikes, and heavy snowfall significantly increase the vulnerability turbine blades to fatigue damage. This susceptibility poses serious risks normal operation longevity turbines, necessitating effective monitoring maintenance strategies. In response these challenges, this paper proposes novel fault detection method specifically designed for analyzing wind blade noise signals. integrates Tyrannosaurus Optimization Algorithm (TROA) with support vector machine (SVM), aiming enhance accuracy reliability detection. The process begins careful preprocessing raw signals collected from during actual operational extracts vital features three key perspectives: time domain, frequency cepstral domain. By constructing comprehensive feature matrix that encapsulates multi-dimensional characteristics, approach ensures all relevant information is captured. Rigorous analysis selection subsequently conducted eliminate redundant data, thereby focusing on retaining most significant classification. A TROA-SVM classification model then developed effectively identify faults blades. performance validated through extensive experiments, which indicate recognition rate 98.7%. higher than traditional methods, SVM, K-Nearest Neighbors (KNN), random forest, demonstrating proposed method’s superiority effectiveness.

Language: Английский

Citations

0
Application of Bayesian networks based on Sequential Monte Carlo simulation and physical model in fault diagnosis of horizontal three-phase separator system DOI
Daqian Liu, Shangfei Song,

Ting Huang

et al.

Ocean Engineering, Journal Year: 2024, Volume and Issue: 306, P. 118139 - 118139

Published: May 13, 2024

Language: Английский

Citations

2
Feature decoupling integrated domain generalization network for bearing fault diagnosis under unknown operating conditions DOI Creative Commons
Qiyang Xiao, Maolin Yang, Jiayuan Yan

et al.

Scientific Reports, Journal Year: 2024, Volume and Issue: 14(1)

Published: Dec. 28, 2024

In real engineering scenarios, the complex and variable operating conditions of mechanical equipment lead to distributional differences between collected fault data training data. This distribution difference can failure deep learning-based diagnostic models. Extracting generalized knowledge from source domain in scenarios where target is not visible key solving this problem. To end, paper, we propose a generalization network for diagnosing bearing faults under unknown conditions, i.e., Feature Decoupled Integrated Domain Generalization Network (FDIDG). First, "feature decoupling" algorithm uncover representations features multiple domains. The aims explore by shrinking domains further generalize reduce coupling conditions. Second, accuracy model improved adopting multi-expert integration strategy decision-making stage utilizing domain-private negative impact edge samples on diagnosis. We conducted several sets cross-domain experiments both public private datasets, results show that FDIDG has excellent capabilities.

Language: Английский

Citations

2
Optimal Time Frequency Fusion Symmetric Dot Pattern Bearing Fault Feature Enhancement and Diagnosis DOI Creative Commons

Guanlong Liang,

Xuewei Song, Zhiqiang Liao

et al.

Sensors, Journal Year: 2024, Volume and Issue: 24(13), P. 4186 - 4186

Published: June 27, 2024

Regarding the difficulty of extracting acquired fault signal features bearings from a strong background noise vibration signal, coupled with fact that one-dimensional (1D) signals provide limited information, an optimal time frequency fusion symmetric dot pattern (SDP) bearing feature enhancement and diagnosis method is proposed. Firstly, are transformed into two-dimensional (2D) by algorithm SDP, which can multi-scale analyze fluctuations at minor scales, as well enhance features. Secondly, bat employed to optimize SDP parameters adaptively. It effectively improve distinctions between various types faults. Finally, model be constructed deep convolutional neural network (DCNN). To validate effectiveness proposed method, Case Western Reserve University's (CWRU) dataset laboratory experimental platform were used. The results illustrate accuracy 100%, proves feasibility method. By comparing other 2D transformer methods, achieves highest in diagnosis. validated superiority methodology.

Language: Английский

Citations

1