[go: up one dir, main page]
More Web Proxy on the site http://driver.im/
You seem to have javascript disabled. Please note that many of the page functionalities won't work as expected without javascript enabled.
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
7.3
3.4
Journals
Sensors
Special Issues
Advanced Sensor Technologies for Fault Diagnosis and Condition Monitoring
sensors-logo
Submit to Special Issue Submit Abstract to Special Issue Review for Sensors Propose a Special Issue

Journal Menu

► Journal Menu

Journal Browser

► Journal Browser
share announcement

Advanced Sensor Technologies for Fault Diagnosis and Condition Monitoring

A special issue of Sensors (ISSN 1424-8220). This special issue belongs to the section "Sensing and Imaging".

Deadline for manuscript submissions: 15 July 2025 | Viewed by 6743

Special Issue Editor

Dr. Md Junayed Hasan

E-Mail Website
Guest Editor
National Subsea Centre, Robert Gordon University, Aberdeen AB21 0BH, UK
Interests: digital condition monitoring; mechanical signal processing; computer vision, machine learning; multimodal data fusion; artificial intelligence; digital fault inspection
Special Issues, Collections and Topics in MDPI journals

Special Issue Information

Dear Colleagues,

In the industrial sector, machinery and mechanical structures are susceptible to deterioration and performance decline over time. Consequently, the collection and processing of data from a variety of sensors has become crucial for the timely diagnosis of deterioration symptoms and the accurate prediction of future health conditions. Using artificial intelligence (AI) technology, models are being developed based on historical sensor data which have enormous potential for fault diagnosis and prognosis in industrial equipment. As the deployment of Internet of Things (IoT) and cloud-based technologies for stateful maintenance increases in the future, AI-powered solutions will become even more crucial for managing the vast quantities of available measurement data for decision making.

This Special Issue aims to investigate fault diagnosis and prognosis of industrial equipment and mechanical structures by utilising a variety of sensors, including those related to image, video, and multimodal information fusion. We welcome researchers to submit articles discussing sensor-based artificial neural network technology, multimodal data fusion, explainable AI solutions, and objects for error diagnosis and prognosis in the context of Industry 4.0, cloud computing, cyber–physical systems, and machine-to-machine interfaces and paradigms.

Dr. Md Junayed Hasan
Guest Editor

Manuscript Submission Information

Manuscripts should be submitted online at www.mdpi.com by registering and logging in to this website. Once you are registered, click here to go to the submission form. Manuscripts can be submitted until the deadline. All submissions that pass pre-check are peer-reviewed. Accepted papers will be published continuously in the journal (as soon as accepted) and will be listed together on the special issue website. Research articles, review articles as well as short communications are invited. For planned papers, a title and short abstract (about 100 words) can be sent to the Editorial Office for announcement on this website.

Submitted manuscripts should not have been published previously, nor be under consideration for publication elsewhere (except conference proceedings papers). All manuscripts are thoroughly refereed through a single-blind peer-review process. A guide for authors and other relevant information for submission of manuscripts is available on the Instructions for Authors page. Sensors is an international peer-reviewed open access semimonthly journal published by MDPI.

Please visit the Instructions for Authors page before submitting a manuscript. The Article Processing Charge (APC) for publication in this open access journal is 2600 CHF (Swiss Francs). Submitted papers should be well formatted and use good English. Authors may use MDPI's English editing service prior to publication or during author revisions.

Keywords

  • multimodal data fusion
  • fault diagnosis
  • digital condition monitoring
  • predictive maintenance
  • artificial intelligence
  • net-zero challenges with AI

Benefits of Publishing in a Special Issue

  • Ease of navigation: Grouping papers by topic helps scholars navigate broad scope journals more efficiently.
  • Greater discoverability: Special Issues support the reach and impact of scientific research. Articles in Special Issues are more discoverable and cited more frequently.
  • Expansion of research network: Special Issues facilitate connections among authors, fostering scientific collaborations.
  • External promotion: Articles in Special Issues are often promoted through the journal's social media, increasing their visibility.
  • e-Book format: Special Issues with more than 10 articles can be published as dedicated e-books, ensuring wide and rapid dissemination.

Further information on MDPI's Special Issue polices can be found here.

Published Papers (5 papers)

Download All Papers
Order results
Result details
Show export options expand_more Show export options expand_less
Select all
Export citation of selected articles as:

Research

20 pages, 8443 KiB  
Article
Damage Detection and Identification on Elevator Systems Using Deep Learning Algorithms and Multibody Dynamics Models
by Josef Koutsoupakis, Dimitrios Giagopoulos, Panagiotis Seventekidis, Georgios Karyofyllas and Amalia Giannakoula
Sensors 2025, 25(1), 101; https://doi.org/10.3390/s25010101 - 27 Dec 2024
Viewed by 193
Abstract
Timely damage detection on a mechanical system can prevent the appearance of catastrophic damage in it, as well as allow for better scheduling of its maintenance and repair process. For this purpose, multiple signal analysis methods have been developed to help identify anomalies [...] Read more.
Timely damage detection on a mechanical system can prevent the appearance of catastrophic damage in it, as well as allow for better scheduling of its maintenance and repair process. For this purpose, multiple signal analysis methods have been developed to help identify anomalies in a system, through quantities such as vibrations or deformations in its critical components. In most applications, however, these data may be scarce or inexistent, hindering the overall process. For this purpose, a novel approach for damage detection and identification on elevator systems is developed in this work, where vibration data obtained through physical measurements and high-fidelity multibody dynamics models are combined with deep learning algorithms. High-quality training data are first generated through multibody dynamics simulations and are then combined with healthy state vibration measurements to train an ensemble of autoencoders and convolutional neural networks for damage detection and classification. A dedicated data acquisition system is then developed and integrated with an elevator cabin, allowing for condition monitoring through this novel methodology. The results indicate that the developed framework can accurately identify damages in the system, hinting at its potential as a powerful structural health monitoring tool for such applications, where manual damage localization would otherwise be considerably time-consuming. Full article
(This article belongs to the Special Issue Advanced Sensor Technologies for Fault Diagnosis and Condition Monitoring)
►▼ Show Figures

Figure 1

Figure 1
<p>Kleemann test tower (<b>a</b>) and experiment floors (<b>b</b>).</p> Full article ">Figure 2
<p>Experimental elevator and subsystems. Subsystems (1)–(3) denote the elevator cabin, chassis, and doors and subsystem (4) denotes the building floor door.</p> Full article ">Figure 3
<p>Elevator cabin (<b>a</b>) and floor (<b>b</b>) door sliding mechanisms.</p> Full article ">Figure 4
<p>DAQ system—Acceleration and proximity sensor placement.</p> Full article ">Figure 5
<p>DAQ and measurement storage protocol—doors opening.</p> Full article ">Figure 6
<p>Healthy elevator acceleration response in the time (<b>left</b>) and frequency (<b>right</b>) domain.</p> Full article ">Figure 7
<p>DAQ system GUI.</p> Full article ">Figure 8
<p>Artificial damage cases on cabin and floor door rails.</p> Full article ">Figure 9
<p>Elevator system MBD model—healthy state.</p> Full article ">Figure 10
<p>Elevator system MBD model—damaged states: (<b>a</b>) Cabin lower rail, (<b>b</b>) Floor lower rail, (<b>c</b>) Cabin and Floor lower rail, and (<b>d</b>) Cabin upper rail.</p> Full article ">Figure 11
<p>Autoencoder-based damage detection framework.</p> Full article ">Figure 12
<p>Damage detection autoencoder architecture.</p> Full article ">Figure 13
<p>Health state classification CNN architecture.</p> Full article ">Figure 14
<p>Healthy state experimental and MBD model system response in the frequency domain.</p> Full article ">Figure 15
<p>Comparison between the experimental and MBD model frequency response data for damage cases 1–6 at Y axis, doors opening.</p> Full article ">Figure 16
<p>DL-SHM framework prediction results—Confusion matrix.</p> Full article ">Figure 17
<p>DL-SHM framework prediction results after additional training with physical measurements—Confusion matrix.</p> Full article ">
24 pages, 19686 KiB  
Article
Utilizing Deep Learning for Defect Inspection in Hand Tool Assembly
by Hong-Dar Lin, Cheng-Kai Jheng, Chou-Hsien Lin and Hung-Tso Chang
Sensors 2024, 24(11), 3635; https://doi.org/10.3390/s24113635 - 4 Jun 2024
Viewed by 1169
Abstract
The integrity of product assembly in the precision assembly industry significantly influences the quality of the final products. During the assembly process, products may acquire assembly defects due to personnel oversight. A severe assembly defect could impair the product’s normal function and potentially [...] Read more.
The integrity of product assembly in the precision assembly industry significantly influences the quality of the final products. During the assembly process, products may acquire assembly defects due to personnel oversight. A severe assembly defect could impair the product’s normal function and potentially cause loss of life or property for the user. For workpiece defect inspection, there is limited discussion on the simultaneous detection of the primary kinds of assembly anomaly (missing parts, misplaced parts, foreign objects, and extra parts). However, these assembly anomalies account for most customer complaints in the traditional hand tool industry. This is because no equipment can comprehensively inspect major assembly defects, and inspections rely solely on professionals using simple tools and their own experience. Thus, this study proposes an automated visual inspection system to achieve defect inspection in hand tool assembly. This study samples the work-in-process from three assembly stations in the ratchet wrench assembly process; an investigation of 28 common assembly defect types is presented, covering the 4 kinds of assembly anomaly in the assembly operation; also, this study captures sample images of various assembly defects for the experiments. First, the captured images are filtered to eliminate surface reflection noise from the workpiece; then, a circular mask is given at the assembly position to extract the ROI area; next, the filtered ROI images are used to create a defect-type label set using manual annotation; after this, the R-CNN series network models are applied to object feature extraction and classification; finally, they are compared with other object detection models to identify which inspection model has the better performance. The experimental results show that, if each station uses the best model for defect inspection, it can effectively detect and classify defects. The average defect detection rate (1-β) of each station is 92.64%, the average misjudgment rate (α) is 6.68%, and the average correct classification rate (CR) is 88.03%. Full article
(This article belongs to the Special Issue Advanced Sensor Technologies for Fault Diagnosis and Condition Monitoring)
►▼ Show Figures

Figure 1

Figure 1
<p>Internal structure of 1/2″ 36T ratchet wrench: (<b>a</b>) physical drawing; (<b>b</b>) exploded view; (<b>c</b>) parts list of assembly parts of the workpiece.</p> Full article ">Figure 2
<p>An assembly diagram of the parts required for each assembly station of the ratchet wrench assembly process.</p> Full article ">Figure 3
<p>Summary chart of the relationship between assembly anomaly kinds and corresponding defect types of ratchet wrenches.</p> Full article ">Figure 4
<p>Example images of four kinds of assembly anomalies and corresponding defect types at the first assembly station of the ratchet wrench.</p> Full article ">Figure 5
<p>Experimental hardware setup for image capture: a photograph and the corresponding schematic diagram.</p> Full article ">Figure 6
<p>Preprocessed images with five mask radii used in this study (unit: pixel).</p> Full article ">Figure 7
<p>Schematic diagram of a network architecture training procedure using R-CNN model.</p> Full article ">Figure 8
<p>Schematic diagram of bounding box regression principle (three colored dots representing the center points of the corresponding bounding boxes).</p> Full article ">Figure 9
<p>Correct and possible failure results of bounding box regression operations.</p> Full article ">Figure 10
<p>Test procedure for ratchet wrench inspection and identification system using R-CNN network models at the first assembly station.</p> Full article ">Figure 11
<p>Schematic diagram of fast R-CNN network architecture.</p> Full article ">Figure 12
<p>Schematic diagram of faster R-CNN network architecture in this study.</p> Full article ">Figure 13
<p>Schematic diagram of Mask R-CNN network architecture in this study.</p> Full article ">Figure 14
<p>Some detection results of sample experiments.</p> Full article ">Figure 15
<p>Schematic diagram of various placement directions of the workpieces in this study.</p> Full article ">Figure 16
<p>Correct classification rate line chart of ratchet wrenches at various placement angles.</p> Full article ">Figure 17
<p>Comparative pictures of the workpieces in the first assembly station after being coated with different lubricant levels (red circles) and then preprocessed and masked.</p> Full article ">Figure 18
<p>Schematic diagram of the operation of the ratchet wrench dynamic visual inspection system.</p> Full article ">Figure 19
<p>Hardware configuration diagram of ratchet wrench dynamic visual detection system.</p> Full article ">Figure 20
<p>Original and preprocessed images captured at different conveyor speed settings in the dynamic visual inspection system.</p> Full article ">Figure 21
<p>ROC plot of ratchet wrench dynamic visual inspection system under different conveying speed settings.</p> Full article ">Figure 22
<p>Correct classification rate curve of ratchet wrench dynamic visual inspection system under different conveying speed settings.</p> Full article ">
19 pages, 7702 KiB  
Article
Feature Extraction of Lubricating Oil Debris Signal Based on Segmentation Entropy with an Adaptive Threshold
by Baojun Yang, Wei Liu, Sheng Lu and Jiufei Luo
Sensors 2024, 24(5), 1380; https://doi.org/10.3390/s24051380 - 21 Feb 2024
Cited by 1 | Viewed by 1042
Abstract
Ferromagnetic debris in lubricating oil, serving as an important communication carrier, can effectively reflect the wear condition of mechanical equipment and predict the remaining useful life. In practice application, the detection signals collected by using inductive sensors contain not only debris signals but [...] Read more.
Ferromagnetic debris in lubricating oil, serving as an important communication carrier, can effectively reflect the wear condition of mechanical equipment and predict the remaining useful life. In practice application, the detection signals collected by using inductive sensors contain not only debris signals but also noise terms, and weak debris features are prone to be distorted, which makes it a severe challenge to debris signature identification and quantitative estimation. In this paper, a debris signature extraction method established on segmentation entropy with an adaptive threshold was proposed, based on which five identification indicators were investigated to improve detection accuracy. The results of the simulations and oil experiment show that the proposed algorithm can effectively identify wear particles and preserve debris signatures. Full article
(This article belongs to the Special Issue Advanced Sensor Technologies for Fault Diagnosis and Condition Monitoring)
►▼ Show Figures

Figure 1

Figure 1
<p>Flowchart of the proposed algorithm.</p> Full article ">Figure 2
<p>The output signal of debris sensor.</p> Full article ">Figure 3
<p>The principle determination of adaptive threshold.</p> Full article ">Figure 4
<p>The thresholds as a function of <span class="html-italic">c</span> computed by numeric method and the approximation solution errors. (<b>a</b>) The solutions computed by numeric method and the analytic solution computed by Equation (5). (<b>b</b>) The errors between the exact solution and the approximation solution.</p> Full article ">Figure 5
<p><span class="html-italic">c</span> for 500 independent trails.</p> Full article ">Figure 6
<p>Feature extraction by the proposed algorithm. (<b>a</b>) Simulated signal; (<b>b</b>) signal after low-pass filtering; (<b>c</b>) signal after harmonics rejection; (<b>d</b>) the normalized segmentation entropy and adaptive threshold; (<b>e</b>) segmentation and identification results.</p> Full article ">Figure 6 Cont.
<p>Feature extraction by the proposed algorithm. (<b>a</b>) Simulated signal; (<b>b</b>) signal after low-pass filtering; (<b>c</b>) signal after harmonics rejection; (<b>d</b>) the normalized segmentation entropy and adaptive threshold; (<b>e</b>) segmentation and identification results.</p> Full article ">Figure 7
<p>Identification results by the feature indicators and estimated amplitude (EA).</p> Full article ">Figure 8
<p>The main structure of experimental platform.</p> Full article ">Figure 9
<p>Signal processing results. (<b>a</b>) Sampled signal; (<b>b</b>) signal after low-pass filtering; (<b>c</b>) signal after harmonics rejection; (<b>d</b>) the normalized segmentation entropy and adaptive threshold; (<b>e</b>) segmentation and identification results.</p> Full article ">Figure 9 Cont.
<p>Signal processing results. (<b>a</b>) Sampled signal; (<b>b</b>) signal after low-pass filtering; (<b>c</b>) signal after harmonics rejection; (<b>d</b>) the normalized segmentation entropy and adaptive threshold; (<b>e</b>) segmentation and identification results.</p> Full article ">Figure 10
<p>Residual noise blocks.</p> Full article ">Figure 11
<p>Captured debris signals.</p> Full article ">Figure 12
<p>Filtering results of the symplectic geometry mode decomposition in [<a href="#B23-sensors-24-01380" class="html-bibr">23</a>].</p> Full article ">Figure 13
<p>Filtering results of the fractional calculus in [<a href="#B16-sensors-24-01380" class="html-bibr">16</a>].</p> Full article ">Figure 14
<p>Filtering results of TIWT with <span class="html-italic">d<sub>l</sub></span> = 4 in [<a href="#B24-sensors-24-01380" class="html-bibr">24</a>].</p> Full article ">Figure 15
<p>Filtering results of TIWT with <span class="html-italic">d<sub>l</sub></span> = 5 in [<a href="#B24-sensors-24-01380" class="html-bibr">24</a>].</p> Full article ">Figure 16
<p>Filtering results of TIWT with <span class="html-italic">d<sub>l</sub></span> = 6 in [<a href="#B24-sensors-24-01380" class="html-bibr">24</a>].</p> Full article ">
30 pages, 3401 KiB  
Article
Adaptive DBSCAN Clustering and GASA Optimization for Underdetermined Mixing Matrix Estimation in Fault Diagnosis of Reciprocating Compressors
by Yanyang Li, Jindong Wang, Haiyang Zhao, Chang Wang and Qi Shao
Sensors 2024, 24(1), 167; https://doi.org/10.3390/s24010167 - 27 Dec 2023
Cited by 2 | Viewed by 1488
Abstract
Underdetermined blind source separation (UBSS) has garnered significant attention in recent years due to its ability to separate source signals without prior knowledge, even when sensors are limited. To accurately estimate the mixed matrix, various clustering algorithms are typically employed to enhance the [...] Read more.
Underdetermined blind source separation (UBSS) has garnered significant attention in recent years due to its ability to separate source signals without prior knowledge, even when sensors are limited. To accurately estimate the mixed matrix, various clustering algorithms are typically employed to enhance the sparsity of the mixed matrix. Traditional clustering methods require prior knowledge of the number of direct signal sources, while modern artificial intelligence optimization algorithms are sensitive to outliers, which can affect accuracy. To address these challenges, we propose a novel approach called the Genetic Simulated Annealing Optimization (GASA) method with Adaptive Density-Based Spatial Clustering of Applications with Noise (DBSCAN) clustering as initialization, named the CYYM method. This approach incorporates two key components: an Adaptive DBSCAN to discard noise points and identify the number of source signals and GASA optimization for automatic cluster center determination. GASA combines the global spatial search capabilities of a genetic algorithm (GA) with the local search abilities of a simulated annealing algorithm (SA). Signal simulations and experimental analysis of compressor fault signals demonstrate that the CYYM method can accurately calculate the mixing matrix, facilitating successful source signal recovery. Subsequently, we analyze the recovered signals using the Refined Composite Multiscale Fuzzy Entropy (RCMFE), which, in turn, enables effective compressor connecting rod fault diagnosis. This research provides a promising approach for underdetermined source separation and offers practical applications in fault diagnosis and other fields. Full article
(This article belongs to the Special Issue Advanced Sensor Technologies for Fault Diagnosis and Condition Monitoring)
►▼ Show Figures

Figure 1

Figure 1
<p>Mixed-signal scatter plot: (<b>a</b>) in the time domain; (<b>b</b>) in the time–frequency domain.</p> Full article ">Figure 2
<p>Time–frequency scatter plot: (<b>a</b>) After the elimination of low energy points. (<b>b</b>) After the detection of single-source points.</p> Full article ">Figure 3
<p>Clustering effect: (<b>a</b>) clustering by DBSCAN; (<b>b</b>) clustering by adaptive DBSCAN.</p> Full article ">Figure 4
<p>Process of adaptive DBSCAN clustering.</p> Full article ">Figure 5
<p>Tree coding structure.</p> Full article ">Figure 6
<p>Two leaf nodes of a tree mutually exchanged: (<b>a</b>) same tree exchange; (<b>b</b>) different tree exchange.</p> Full article ">Figure 7
<p>Weight coefficient decision diagram. (<b>a</b>) the trend graph of the fitness function as the power exponent increases; (<b>b</b>) the trend graph of computation time with an increasing power exponent.</p> Full article ">Figure 8
<p>The flowchart of the CYYM method.</p> Full article ">Figure 9
<p>Waveforms of source signals: (<b>a</b>) in the time domain. (<b>b</b>) in the frequency domain.</p> Full article ">Figure 10
<p>Mixed signals: (<b>a</b>) time domain waveforms; (<b>b</b>) envelope spectra.</p> Full article ">Figure 11
<p>(<b>a</b>) Normalized time–frequency scatterplots. (<b>b</b>) Clusted by GASA. (<b>c</b>) Clusted by improved DBSCAN. (<b>d</b>) Clusted by CYYM.</p> Full article ">Figure 12
<p>Time− domain signal comparison diagram: (<b>a</b>) source signals; (<b>b</b>) recovery Signal.</p> Full article ">Figure 13
<p>Frequency domain signal comparison diagram: (<b>a</b>) source signals; (<b>b</b>) recovery signal.</p> Full article ">Figure 14
<p>Source signals: (<b>a</b>) Waveforms. (<b>b</b>) Fourier spectrums.</p> Full article ">Figure 15
<p>Mixed signals: (<b>a</b>) waveforms; (<b>b</b>) Fourier spectra.</p> Full article ">Figure 16
<p>Time-domain signal: (<b>a</b>) Source signal <math display="inline"><semantics> <msub> <mi>s</mi> <mn>1</mn> </msub> </semantics></math>. (<b>b</b>) <math display="inline"><semantics> <msub> <mi>s</mi> <mn>1</mn> </msub> </semantics></math> obtained by TIFROM method.</p> Full article ">Figure 17
<p>Time-domain signal: (<b>a</b>) Source signal <math display="inline"><semantics> <msub> <mi>s</mi> <mn>2</mn> </msub> </semantics></math>. (<b>b</b>) <math display="inline"><semantics> <msub> <mi>s</mi> <mn>2</mn> </msub> </semantics></math> obtained by TIFROM method.</p> Full article ">Figure 18
<p>Time-domain signal: (<b>a</b>) Source signal <math display="inline"><semantics> <msub> <mi>s</mi> <mn>3</mn> </msub> </semantics></math>. (<b>b</b>) <math display="inline"><semantics> <msub> <mi>s</mi> <mn>3</mn> </msub> </semantics></math> obtained by TIFROM method.</p> Full article ">Figure 19
<p>Time -domain signal: (<b>a</b>) Source signal <math display="inline"><semantics> <msub> <mi>s</mi> <mn>4</mn> </msub> </semantics></math>. (<b>b</b>) <math display="inline"><semantics> <msub> <mi>s</mi> <mn>4</mn> </msub> </semantics></math> obtained by TIFROM method.</p> Full article ">Figure 20
<p>Time-domain signal: (<b>a</b>) Source signal <math display="inline"><semantics> <msub> <mi>s</mi> <mn>1</mn> </msub> </semantics></math>. (<b>b</b>) <math display="inline"><semantics> <msub> <mi>s</mi> <mn>1</mn> </msub> </semantics></math> obtained by DEMIX method.</p> Full article ">Figure 21
<p>Time -domain signal: (<b>a</b>) Source signal <math display="inline"><semantics> <msub> <mi>s</mi> <mn>2</mn> </msub> </semantics></math>. (<b>b</b>) <math display="inline"><semantics> <msub> <mi>s</mi> <mn>2</mn> </msub> </semantics></math> obtained by DEMIX method.</p> Full article ">Figure 22
<p>Time-domain signal: (<b>a</b>) Source signal <math display="inline"><semantics> <msub> <mi>s</mi> <mn>3</mn> </msub> </semantics></math>. (<b>b</b>) <math display="inline"><semantics> <msub> <mi>s</mi> <mn>3</mn> </msub> </semantics></math> obtained by DEMIX method.</p> Full article ">Figure 23
<p>Time-domain signal: (<b>a</b>) Source signal <math display="inline"><semantics> <msub> <mi>s</mi> <mn>4</mn> </msub> </semantics></math>. (<b>b</b>) <math display="inline"><semantics> <msub> <mi>s</mi> <mn>4</mn> </msub> </semantics></math> obtained by DEMIX method.</p> Full article ">Figure 24
<p>Number of clusters performance map.</p> Full article ">Figure 25
<p>Time-domain signal: (<b>a</b>) Source signals. (<b>b</b>) Estimated signals obtained by CYYM method.</p> Full article ">Figure 26
<p>DW-10/12-27-Xlll type two-stage double-acting reciprocating compressor.</p> Full article ">Figure 27
<p>The driving schematic of the compressor mechanism.</p> Full article ">Figure 28
<p>Composition of the reciprocating compressor connecting rod: (<b>a</b>) connecting rod; (<b>b</b>) big head of the connecting rod; (<b>c</b>) bearing bush; (<b>d</b>) failure bearing bush.</p> Full article ">Figure 29
<p>Mixed signals: (<b>a</b>) time−domain waveforms; (<b>b</b>) envelope spectra.</p> Full article ">Figure 30
<p>Time-domain contrast diagram of compressor signals: (<b>a</b>) source signals; (<b>b</b>) recovery signals by the CYYM method.</p> Full article ">Figure 31
<p>Frequency domain contrast diagram of compressor signals: (<b>a</b>) source signals; (<b>b</b>) recovery signals by the CYYM method.</p> Full article ">Figure 32
<p>Comparison diagram of correlation coefficients.</p> Full article ">Figure 33
<p>Comparison diagram of NMSE.</p> Full article ">Figure 34
<p>Comparison diagram of SIR.</p> Full article ">Figure 35
<p>RCMFE characterization curve vs. fault library identification plot: (<b>a</b>) normal state; (<b>b</b>) big end failure; (<b>c</b>) small end failure.</p> Full article ">
18 pages, 6557 KiB  
Article
Fault Detection in Power Distribution Networks Based on Comprehensive-YOLOv5
by Shengsuo Niu, Xiaosen Zhou, Dasen Zhou, Zhiyao Yang, Haiping Liang and Haifeng Su
Sensors 2023, 23(14), 6410; https://doi.org/10.3390/s23146410 - 14 Jul 2023
Cited by 7 | Viewed by 2123
Abstract
Real-time fault detection in power distribution networks has become a popular issue in current power systems. However, the low power and computational capabilities of edge devices often fail to meet the requirements of real-time detection. To overcome these challenges, this paper proposes a [...] Read more.
Real-time fault detection in power distribution networks has become a popular issue in current power systems. However, the low power and computational capabilities of edge devices often fail to meet the requirements of real-time detection. To overcome these challenges, this paper proposes a lightweight algorithm, named Comprehensive-YOLOv5, for identifying defects in distribution networks. The proposed method focuses on achieving rapid localization and accurate identification of three common defects: insulator without loop, cable detachment from the insulator, and cable detachment from the spacer. Based on the You Only Look Once version 5 (YOLOv5) algorithm, this paper adopts GhostNet to reconstruct the original backbone of YOLOv5; introduces Bidirectional Feature Pyramid Network (BiFPN) structure to replace Path Aggregation Network (PANet) for feature fusion, which enhances the feature fusion ability; and replaces Generalized Intersection over Union GIOU with Focal Extended Intersection over Union (Focal-EIOU) to optimize the loss function, which improves the mean average precision and speed of the algorithm. The effectiveness of the improved Comprehensive-YOLOv5 algorithm is verified through a “morphological experiment”, while an “algorithm comparison experiment” confirms its superiority over other algorithms. Compared with the original YOLOv5, the Comprehensive-YOLOv5 algorithm improves mean average precision (mAP) from 88.3% to 90.1% and increases Frames per second (FPS) from 20 to 52 frames. This improvement significantly reduces false positives and false negatives in defect detection. Consequently, the proposed algorithm enhances detection speed and improves inspection efficiency, providing a viable solution for real-time detection and deployment at the edge of power distribution networks. Full article
(This article belongs to the Special Issue Advanced Sensor Technologies for Fault Diagnosis and Condition Monitoring)
►▼ Show Figures

Figure 1

Figure 1
<p>Network Model of YOLOv5.</p> Full article ">Figure 2
<p>FPN + PANet structure. (<b>a</b>) FPN backbone; (<b>b</b>) PANet backbone.</p> Full article ">Figure 3
<p>GhostConv module.</p> Full article ">Figure 4
<p>C3Ghost module.</p> Full article ">Figure 5
<p>Schematic diagram of the BiFPN structure.</p> Full article ">Figure 6
<p>Comparison between PANet and BiFPN structures.</p> Full article ">Figure 7
<p>Network Model of Comprehensive-YOLOv5.</p> Full article ">Figure 8
<p>Three Typical Defects in Distribution Grids. (<b>a</b>) Insulator Ring Absence; (<b>b</b>) Cable Detachment from Insulators; (<b>c</b>) Cable Detachment from Spacers.</p> Full article ">Figure 8 Cont.
<p>Three Typical Defects in Distribution Grids. (<b>a</b>) Insulator Ring Absence; (<b>b</b>) Cable Detachment from Insulators; (<b>c</b>) Cable Detachment from Spacers.</p> Full article ">Figure 9
<p>Distribution Grid Defect Detection Process Flowchart.</p> Full article ">Figure 10
<p>Loss graph.</p> Full article ">Figure 11
<p>Comparative Detection Results Chart. (<b>a</b>) YOLOv5 Detection Results Chart. (<b>b</b>) Comprehensive-YOLOv5 Detection Results Chart.</p> Full article ">Figure 11 Cont.
<p>Comparative Detection Results Chart. (<b>a</b>) YOLOv5 Detection Results Chart. (<b>b</b>) Comprehensive-YOLOv5 Detection Results Chart.</p> Full article ">
Show export options expand_more Show export options expand_less
Select all
Export citation of selected articles as:
 
Displaying articles 1-5
Back to TopTop