A Comparative Analysis of Deep Learning Convolutional Neural Network Architectures for Fault Diagnosis of Broken Rotor Bars in Induction Motors DOI Creative Commons
Kevin Barrera, Jordi Burriel-Valencia, Ángel Sapena-Bañó

et al.

Sensors, Journal Year: 2023, Volume and Issue: 23(19), P. 8196 - 8196

Published: Sept. 30, 2023

Induction machines (IMs) play a critical role in various industrial processes but are susceptible to degenerative failures, such as broken rotor bars. Effective diagnostic techniques essential addressing these issues. In this study, we propose the utilization of convolutional neural networks (CNNs) for detection To accomplish this, generated dataset comprising current samples versus angular position using finite element method magnetics (FEMM) software squirrel-cage with 28 bars, including scenarios 0 6 bars at every possible relative position. The consists total 16,050 per motor. We evaluated performance six different CNN architectures, namely Inception V4, NasNETMobile, ResNET152, SeNET154, VGG16, and VGG19. Our automatic classification system demonstrated an impressive 99% accuracy detecting VGG19 performing exceptionally well. Specifically, exhibited high accuracy, precision, recall, F1-Score, values approaching 0.994 0.998. Notably, crucial activations its feature maps, particularly after domain-specific training, highlighting effectiveness fault detection. Comparing architectures assists selecting most suitable one application based on processing time, effectiveness, training losses. This research suggests that deep learning can detect induction comparable traditional methods by analyzing signals CNNs.

Language: Английский

Current Status, Challenges, and Possible Solutions of EEG-Based Brain-Computer Interface: A Comprehensive Review DOI Creative Commons
Mamunur Rashid, Norizam Sulaiman, Anwar P. P. Abdul Majeed

et al.

Frontiers in Neurorobotics, Journal Year: 2020, Volume and Issue: 14

Published: June 3, 2020

Brain-Computer Interface (BCI), in essence, aims at controlling different assistive devices through the utilization of brain waves. It is worth noting that application BCI not limited to medical applications, and hence, research this field has gained due attention. Moreover, significant number related publications over past two decades further indicates consistent improvements breakthroughs have been made particular field. Nonetheless, it also mentioning with these improvements, new challenges are constantly discovered. This article provides a comprehensive review state-of-the-art complete system. First, brief overview electroencephalogram (EEG)-based systems given. Secondly, considerable popular applications reviewed terms electrophysiological control signals, feature extraction, classification algorithms, performance evaluation metrics. Finally, recent discussed, possible solutions mitigate issues recommended.

Language: Английский

Citations

345
Deep learning for motor imagery EEG-based classification: A review DOI
Ali Al-Saegh, Shefa A. Dawwd,

Jassim M. Abdul-Jabbar

et al.

Biomedical Signal Processing and Control, Journal Year: 2020, Volume and Issue: 63, P. 102172 - 102172

Published: Oct. 8, 2020

Language: Английский

Citations

312
Brain functional and effective connectivity based on electroencephalography recordings: A review DOI Creative Commons
Jun Cao, Yifan Zhao, Xiaocai Shan

et al.

Human Brain Mapping, Journal Year: 2021, Volume and Issue: 43(2), P. 860 - 879

Published: Oct. 20, 2021

Functional connectivity and effective of the human brain, representing statistical dependence directed information flow between cortical regions, significantly contribute to study intrinsic brain network its functional mechanism. Many recent studies on electroencephalography (EEG) have been focusing modeling estimating due increasing evidence that it can help better understand various neurological conditions. However, there is a lack comprehensive updated review EEG-based connectivity, particularly visualization options associated machine learning applications, aiming translate those techniques into useful clinical tools. This article reviews undertaken over last few years, in terms estimation, visualization, applications with classifiers. Methods are explored discussed from dimensions, such as either linear or nonlinear, parametric nonparametric, time-based, frequency-based time-frequency-based. Then followed by novel methods, grouped Heat Map, data statistics, Head explore variation across different regions. Finally, current challenges related research roadmap for future presented.

Language: Английский

Citations

143
A fuzzy-enhanced deep learning approach for early detection of Covid-19 pneumonia from portable chest X-ray images DOI
Cosimo Ieracitano, Nadia Mammone, Mario Versaci

et al.

Neurocomputing, Journal Year: 2022, Volume and Issue: 481, P. 202 - 215

Published: Jan. 21, 2022

Language: Английский

Citations

122
LMDA-Net:A lightweight multi-dimensional attention network for general EEG-based brain-computer interfaces and interpretability DOI Creative Commons
Zhengqing Miao,

Meirong Zhao,

Xin Zhang

et al.

NeuroImage, Journal Year: 2023, Volume and Issue: 276, P. 120209 - 120209

Published: June 2, 2023

Electroencephalography (EEG)-based brain-computer interfaces (BCIs) pose a challenge for decoding due to their low spatial resolution and signal-to-noise ratio. Typically, EEG-based recognition of activities states involves the use prior neuroscience knowledge generate quantitative EEG features, which may limit BCI performance. Although neural network-based methods can effectively extract they often encounter issues such as poor generalization across datasets, high predicting volatility, model interpretability. To address these limitations, we propose novel lightweight multi-dimensional attention network, called LMDA-Net. By incorporating two modules designed specifically signals, channel module depth module, LMDA-Net is able integrate features from multiple dimensions, resulting in improved classification performance various tasks. was evaluated on four high-impact public including motor imagery (MI) P300-Speller, compared with other representative models. The experimental results demonstrate that outperforms terms accuracy achieving highest all datasets within 300 training epochs. Ablation experiments further confirm effectiveness module. facilitate an in-depth understanding extracted by LMDA-Net, class-specific network feature interpretability algorithms are suitable evoked responses endogenous activities. mapping output specific layer time or domain through class activation maps, visualizations provide interpretable analysis establish connections time-spatial neuroscience. In summary, shows great potential general

Language: Английский

Citations

48
Cognitive neuroscience and robotics: Advancements and future research directions DOI Creative Commons
Sichao Liu, Lihui Wang, Robert X. Gao

et al.

Robotics and Computer-Integrated Manufacturing, Journal Year: 2023, Volume and Issue: 85, P. 102610 - 102610

Published: July 24, 2023

In recent years, brain-based technologies that capitalise on human abilities to facilitate human–system/robot interactions have been actively explored, especially in brain robotics. Brain–computer interfaces, as applications of this conception, set a path convert neural activities recorded by sensors from the scalp via electroencephalography into valid commands for robot control and task execution. Thanks advancement sensor technologies, non-invasive invasive headsets designed developed achieve stable recording brainwave signals. However, robust accurate extraction interpretation signals robotics are critical reliable task-oriented opportunistic such brainwave-controlled robotic interactions. response need, pervasive advanced analytical approaches translating merging functions, behaviours, tasks, environmental information focus brain-controlled applications. These methods composed signal processing, feature extraction, representation activities, command conversion control. Artificial intelligence algorithms, deep learning, used classification, recognition, identification patterns intent underlying brainwaves form electroencephalography. Within context, paper provides comprehensive review past current status at intersection robotics, neuroscience, artificial highlights future research directions.

Language: Английский

Citations

46
A novel explainable machine learning approach for EEG-based brain-computer interface systems DOI
Cosimo Ieracitano, Nadia Mammone, Amir Hussain

et al.

Neural Computing and Applications, Journal Year: 2021, Volume and Issue: 34(14), P. 11347 - 11360

Published: March 6, 2021

Language: Английский

Citations

91
Deep Learning in EEG: Advance of the Last Ten-Year Critical Period DOI Creative Commons
Shu Gong,

Kaibo Xing,

Andrzej Cichocki

et al.

IEEE Transactions on Cognitive and Developmental Systems, Journal Year: 2021, Volume and Issue: 14(2), P. 348 - 365

Published: May 13, 2021

Deep learning has achieved excellent performance in a wide range of domains, especially speech recognition and computer vision. Relatively less work been done for EEG, but there is still significant progress attained the last decade. Due to lack comprehensive topic widely covered survey deep we attempt summarize recent provide an overview, as well perspectives future developments. We first briefly mention artifacts removal EEG signal then introduce models that have utilized processing classification. Subsequently, applications are reviewed by categorizing them into groups such brain-computer interface, disease detection, emotion recognition. They followed discussion, which pros cons presented directions challenges proposed. hope this paper could serve summary past beginning further developments achievements studies based on learning.

Language: Английский

Citations

76
AutoEncoder Filter Bank Common Spatial Patterns to Decode Motor Imagery From EEG DOI
Nadia Mammone, Cosimo Ieracitano, Hojjat Adeli

et al.

IEEE Journal of Biomedical and Health Informatics, Journal Year: 2023, Volume and Issue: 27(5), P. 2365 - 2376

Published: Feb. 9, 2023

The present paper introduces a novel method, named AutoEncoder-Filter Bank Common Spatial Patterns (AE-FBCSP), to decode imagined movements from electroencephalography (EEG). AE-FBCSP is an extension of the well-established FBCSP and based on global (cross-subject) subsequent transfer learning subject-specific (intra-subject) approach. A multi-way also introduced in this paper. Features are extracted high-density EEG (64 electrodes), by means FBCSP, used train custom AE, unsupervised way, project features into compressed latent space. Latent supervised classifier (feed forward neural network) movement. proposed method was tested using public dataset EEGs collected 109 subjects. consists right-hand, left-hand, both hands, feet motor imagery resting EEGs. extensively 3-way classification (right hand vs left resting) 2-way, 4-way 5-way ones, cross- intra-subject analysis. outperformed standard statistically significant way (p > 0.05) achieved average accuracy 89.09% classification. methodology performed better than other comparable methods literature, applied same dataset, tasks. One most interesting outcomes that remarkably increased number subjects responded with very high accuracy, which fundamental requirement for BCI systems be practice.

Language: Английский

Citations

38
Detection of Atrial Fibrillation Using a Machine Learning Approach DOI Creative Commons
Sidrah Liaqat, Kia Dashtipour, Adnan Zahid

et al.

Information, Journal Year: 2020, Volume and Issue: 11(12), P. 549 - 549

Published: Nov. 26, 2020

The atrial fibrillation (AF) is one of the most well-known cardiac arrhythmias in clinical practice, with a prevalence 1–2% community, which can increase risk stroke and myocardial infarction. detection AF electrocardiogram (ECG) improve early diagnosis. In this paper, we have further developed framework for processing ECG signal order to determine episodes. We implemented machine learning deep algorithms detect AF. Moreover, experimental results show that better performance be achieved long short-term memory (LSTM) as compared other algorithms. initial illustrate algorithms, such LSTM convolutional neural network (CNN), (10%) classifiers, support vectors, logistic regression, etc. This preliminary work help clinicians high accuracy less probability errors, ultimately result reduction fatality rate.

Language: Английский

Citations

61