Few-shot bearing fault diagnosis method based on an EEMD parallel neural network and a relation network DOI Creative Commons
Cunsheng Zhao, Bo Tong, Chao Zhou

et al.

Advances in Mechanical Engineering, Journal Year: 2024, Volume and Issue: 16(10)

Published: Oct. 1, 2024

Bearing fault diagnosis presents challenges, such as insufficient samples and significant data distribution variation in different bearing operating conditions. These problems cause traditional deep learning models to show poor generality accuracy during diagnosis. To address these this paper proposed a few-shot method based on an Ensemble Empirical Mode Decomposition (EEMD) parallel neural network relation (RN). First, the original vibration signal was decomposed by EEMD, while components were processed via Short Time Fourier Transfor (STFT) obtain two-dimensional time-frequency feature map. Then, used for initial extraction, after which extracted features fused construct more accurate multi-dimensional A precise vector generated embedding module of RN, support query sets stitched create set. Finally, RN nonlinear distance determination set generate score variable condition In paper, EEMD is introduced into characteristics signal. Original decomposition, STFT transformation splicing effectively improve randomness blindness convolution operations, extraction thus overall diagnostic performance model. The experimental results showed that model obtained higher than matching (MN) meta-learning (MLFD) methods. 5Way-Nshot >80%, 5Way-10shot highest at 95.2%.

Language: Английский

An adaptive feature mode decomposition based on a novel health indicator for bearing fault diagnosis DOI
Sumika Chauhan, Govind Vashishtha, Rajesh Kumar

et al.

Measurement, Journal Year: 2024, Volume and Issue: 226, P. 114191 - 114191

Published: Jan. 21, 2024

Language: Английский

Citations

41
Designing of optimal digital IIR filter in the multi-objective framework using an evolutionary algorithm DOI
Sumika Chauhan, Manmohan Singh, Ashwani Kumar Aggarwal

et al.

Engineering Applications of Artificial Intelligence, Journal Year: 2023, Volume and Issue: 119, P. 105803 - 105803

Published: Jan. 3, 2023

Language: Английский

Citations

39
Multi-scale ensemble dispersion Lempel-Ziv complexity and its application on feature extraction for ship-radiated noise DOI
Yuxing Li, Yuhan Zhou, Shangbin Jiao

et al.

Applied Acoustics, Journal Year: 2024, Volume and Issue: 218, P. 109890 - 109890

Published: Jan. 29, 2024

Language: Английский

Citations

13
Advanced CKD detection through optimized metaheuristic modeling in healthcare informatics DOI Creative Commons
Anas Bilal, Abdulkareem Alzahrani, Abdullah Almuhaimeed

et al.

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

Published: June 1, 2024

Abstract Data categorization is a top concern in medical data to predict and detect illnesses; thus, it applied modern healthcare informatics. In informatics, machine learning deep models have enjoyed great attention for categorizing improving illness detection. However, the existing techniques, such as features with high dimensionality, computational complexity, long-term execution duration, raise fundamental problems. This study presents novel classification model employing metaheuristic methods maximize efficient positives on Chronic Kidney Disease diagnosis. The initially massively pre-processed, where purified various mechanisms, including missing values resolution, transformation, employment of normalization procedures. focus processes leverage handling prepare analysis. We adopt Binary Grey Wolf Optimization method, reliable subset selection feature using metaheuristics. operation aimed at prediction accuracy. step, adopts Extreme Learning Machine hidden nodes through optimization presence CKD. complete classifier evaluation employs established measures, recall, specificity, kappa, F-score, accuracy, addition selection. related show that proposed approach records levels which better than models.

Language: Английский

Citations

11
An adaptive metaheuristic optimization approach for Tennessee Eastman process for an industrial fault tolerant control system DOI Creative Commons
Faizan E Mustafa, Ijaz Ahmed, Abdul Basit

et al.

PLoS ONE, Journal Year: 2024, Volume and Issue: 19(2), P. e0296471 - e0296471

Published: Feb. 21, 2024

The Tennessee Eastman Process (TEP) is widely recognized as a standard reference for assessing the effectiveness of fault detection and false alarm tracking methods in intricate industrial operations. This paper presents novel methodology that employs Adaptive Crow Search Algorithm (ACSA) to improve identification capabilities mitigate occurrence alarms TEP. ACSA an optimization approach draws inspiration from observed behavior crows their natural environment. algorithm possesses capability adapt its search response changing dynamics process. primary objective our research devise monitoring strategy adaptable nature, with aim efficiently identifying faults within TEP while simultaneously minimizing alarms. applied order enhance variables, thresholds, decision criteria selection configuration. When compared traditional static approaches, ACSA-based better at finding reducing because it adapts well changes process disturbances. In assess efficacy suggested methodology, we have conducted comprehensive simulations on dataset. findings suggest based demonstrates superior rates concurrently mitigating frequency addition, flexibility allows manage variations, disturbances, uncertainties, thereby enhancing robustness reliability practical scenarios. To validate proposed approach, extensive were results indicate achieves higher Moreover, adaptability enables effectively handle making robust reliable real-world applications. contributions this extend beyond TEP, adaptive utilizing can be other complex processes. study provide valuable insights into development advanced techniques, offering significant benefits terms safety, reliability, operational efficiency.

Language: Английский

Citations

10
Optimal filter design using mountain gazelle optimizer driven by novel sparsity index and its application to fault diagnosis DOI
Sumika Chauhan, Govind Vashishtha, Radosław Zimroz

et al.

Applied Acoustics, Journal Year: 2024, Volume and Issue: 225, P. 110200 - 110200

Published: July 30, 2024

Language: Английский

Citations

9
Optimization of spectral kurtosis-based filtering through flow direction algorithm for early fault detection DOI
Govind Vashishtha, Sumika Chauhan, Radosław Zimroz

et al.

Measurement, Journal Year: 2024, Volume and Issue: unknown, P. 115737 - 115737

Published: Sept. 1, 2024

Language: Английский

Citations

8
Unsupervised Learning Model of Sparse Filtering Enhanced Using Wasserstein Distance for Intelligent Fault Diagnosis DOI
Govind Vashishtha, Rajesh Kumar

Journal of Vibration Engineering & Technologies, Journal Year: 2022, Volume and Issue: 11(7), P. 2985 - 3002

Published: Oct. 10, 2022

Language: Английский

Citations

23
Data augmentation on fault diagnosis of wind turbine gearboxes with an enhanced flow-based generative model DOI
Wenliao Du, Pengxiang Zhu, Ziqiang Pu

et al.

Measurement, Journal Year: 2023, Volume and Issue: 225, P. 113985 - 113985

Published: Dec. 6, 2023

Language: Английский

Citations

14
Research on Remaining Useful Life Prediction of Bearings Based on MBCNN-BiLSTM DOI Creative Commons

Jian Li,

Faguo Huang, Haihua Qin

et al.

Applied Sciences, Journal Year: 2023, Volume and Issue: 13(13), P. 7706 - 7706

Published: June 29, 2023

For safe maintenance and to reduce the risk of mechanical faults, remaining useful life (RUL) estimate bearings is significant. The typical methods bearings’ RUL prediction suffer from low accuracy because difficulty in extracting features. With aim improving prediction, an approach based on multi-branch improved convolutional network (MBCNN) with global attention mechanism combined bi-directional long- short-term memory (BiLSTM) proposed for prediction. Firstly, original vibration signal fast Fourier transformed obtain frequency domain then normalized. Secondly, are input into designed MBCNN as two branches extract spatial features, BiLSTM further timing mapped by fully connected achieve purpose Finally, example validation was performed a publicly available bearing degradation dataset. Compared some existing methods, mean absolute root square errors predictions were reduced “22.2%” “50.0%” “26.1%” “52.8%”, respectively, which proved effectiveness feasibility method.

Language: Английский

Citations

12