Mei Wang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Mei Wang has a total of 58 co-authors. Here are the top co-authors by number of shared publications:
- Xia Fang
School of Mechanical Engineering, Sichuan University, Chengdu, China
4 shared publications - Jie Wang
Sichuan University
3 shared publications - Jigang Huang
School of Mechanical Engineering, Sichuan University, Chengdu 610065, China
3 shared publications - Zhigao Wang
State Grid Sichuan Electric Power Research Institute, Chengdu 610041, China
3 shared publications - Jie Ren
National Key Laboratory of Digital and Agile Aircraft Design, AVIC Chengdu Aircraft Design and Research Institute, Chengdu 610000, China
2 shared publications - Junjie Chen
State Grid Sichuan Electric Power Research Institute, Chengdu 610041, China
2 shared publications
Latest Publications
- Open Access
- 5 November 2026
- Published by MDPI AG in Electronics
- Vol. 15 (21) , 5142
Decoupled Candidate Detection and Multi-Evidence Verification for Tiny Surface Defect Inspection
Tiny surface defects are difficult to inspect because of their small scale, weak contrast, diverse morphology, and similarity to normal textures. To address the conflict between early candidate retention and false-positive suppression, this study proposes a coarse-to-fine inspection framework that separates candidate detection from final verification. You Only Look Once (YOLO) 26s first generates defect candidates, which are locally reconstructed and analyzed by a region-of-interest (ROI)-guided Multi-Task SAM (ROI-MT-SAM) for joint segmentation, semantic prediction, and mask-quality estimation. Multi-Evidence Verification (MEV) then integrates complementary confidence and spatial cues to refine the final decision, while residual-transfer augmentation improves defect diversity under limited-data conditions. Experiments show that ROI-MT-SAM improves mean Intersection over Union (mIoU) from 0.3918 to 0.8869 compared with whole-image SAM, with particularly substantial gains for small defects. MEV further reduces false positives from 245 to 46 and achieves a precision of 0.9565 and an F1-score of 0.9241. These results demonstrate the potential of combining candidate-centered representation with multi-evidence verification for tiny-defect inspection of automotive interior surfaces under the evaluated conditions.
- 5 November 2026
- Published by MDPI AG in Electronics
- Vol. 15 (21) , 5142
Open AccessDecoupled Candidate Detection and Multi-Evidence Verification for Tiny Surface Defect Inspection
Tiny surface defects are difficult to inspect because of their small scale, weak contrast, diverse morphology, and similarity to normal textures. To address the conflict between early candidate retention and false-positive suppression, this study proposes a coarse-to-fine inspection framework that separates candidate detection from final verification. You Only Look Once (YOLO) 26s first generates defect candidates, which are locally reconstructed and analyzed by a region-of-interest (ROI)-guided Multi-Task SAM (ROI-MT-SAM) for joint segmentation, semantic prediction, and mask-quality estimation. Multi-Evidence Verification (MEV) then integrates complementary confidence and spatial cues to refine the final decision, while residual-transfer augmentation improves defect diversity under limited-data conditions. Experiments show that ROI-MT-SAM improves mean Intersection over Union (mIoU) from 0.3918 to 0.8869 compared with whole-image SAM, with particularly substantial gains for small defects. MEV further reduces false positives from 245 to 46 and achieves a precision of 0.9565 and an F1-score of 0.9241. These results demonstrate the potential of combining candidate-centered representation with multi-evidence verification for tiny-defect inspection of automotive interior surfaces under the evaluated conditions.
- Open Access
- 28 October 2026
- Published by MDPI AG in Applied Sciences
- Vol. 16 (21) , 10686
Acoustic-Array Diagnosis of Coupled Planetary Gearbox Faults Under Controlled 3D Spatial Interference with Physically Motivated Structural Priors
Non-contact acoustic arrays offer a complementary route for planetary-gearbox monitoring, but multiple sound sources introduce spatial ambiguity and masking. We propose a multi-modal mixture-of-experts framework (MM-MoE) with physically motivated structural priors for 16-channel acoustic signals. Four learned prototype waveforms replace per-channel wavelet transforms, and an adapted Vision Mamba encoder produces channel-aligned spectral embeddings. A waveform branch models temporal patterns, a coupling prior learned from inter-channel spectral similarity biases cross-modal attention, and task-aware expert routing supports simultaneous scene parsing and fault diagnosis. Test-bench gearbox recordings were replayed through two loudspeakers at controlled positions, heights and sound-pressure levels in a semi-anechoic chamber. Over five training runs, MM-MoE achieves window-level accuracies of 89.4% for 16-class spatial-energy scene parsing and 76.5% for 16-class coupled fault diagnosis, with 69.2% joint exact match (JEM), 13.4 percentage points above the strongest evaluated baseline (95% confidence interval: 12.4–14.4). Aggregating the windows of each recording raises JEM to 87.1%. When source regions are excluded from training, ordered fault accuracy on them is 68.3%, compared with 75.5% on seen regions. The findings are limited to controlled loudspeaker replay; validation on operating machinery remains necessary.
- 28 October 2026
- Published by MDPI AG in Applied Sciences
- Vol. 16 (21) , 10686
Open AccessAcoustic-Array Diagnosis of Coupled Planetary Gearbox Faults Under Controlled 3D Spatial Interference with Physically Motivated Structural Priors
Non-contact acoustic arrays offer a complementary route for planetary-gearbox monitoring, but multiple sound sources introduce spatial ambiguity and masking. We propose a multi-modal mixture-of-experts framework (MM-MoE) with physically motivated structural priors for 16-channel acoustic signals. Four learned prototype waveforms replace per-channel wavelet transforms, and an adapted Vision Mamba encoder produces channel-aligned spectral embeddings. A waveform branch models temporal patterns, a coupling prior learned from inter-channel spectral similarity biases cross-modal attention, and task-aware expert routing supports simultaneous scene parsing and fault diagnosis. Test-bench gearbox recordings were replayed through two loudspeakers at controlled positions, heights and sound-pressure levels in a semi-anechoic chamber. Over five training runs, MM-MoE achieves window-level accuracies of 89.4% for 16-class spatial-energy scene parsing and 76.5% for 16-class coupled fault diagnosis, with 69.2% joint exact match (JEM), 13.4 percentage points above the strongest evaluated baseline (95% confidence interval: 12.4–14.4). Aggregating the windows of each recording raises JEM to 87.1%. When source regions are excluded from training, ordered fault accuracy on them is 68.3%, compared with 75.5% on seen regions. The findings are limited to controlled loudspeaker replay; validation on operating machinery remains necessary.
- Open Access
Synthetic-Data-Augmented Corrosion-Severity Grading of Grounding Connectors: A Colorimetric Benchmark and Kinetics-Aware Ranking
Corrosion-severity grading of grounding-grid connectors from optical images supports proactive power-infrastructure maintenance. Existing approaches rely on single-time-point, manually thresholded hue–saturation–value (HSV) metrics and static multi-criteria decision-making (MCDM) frameworks that cannot capture corrosion dynamics. In this paper we present a pipeline that (1) defines a four-class corrosion grade from an HSV area fraction (Scorr) measured on RGBA optical images, and validates those labels against a baseline-referenced CIEDE2000 metric zero-referenced to each connector’s as-received appearance; (2) generates 240 color-prior-constrained procedural synthetic images from 53 real images across six connector types; (3) fine-tunes a ResNet-18 to estimate corrosion coverage continuously, deriving the reported severity class from that estimate rather than predicting it directly; and (4) fits power-law kinetics C(t) = k·tn to the Scorr time series, propagates bootstrap uncertainty into a Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) framework, and reports kinetics-aware rankings as rank probabilities. The label validation quantifies two limitations of single-threshold HSV grading: a material-color offset that scores an unexposed copper connector at Scorr = 0.442, and insensitivity to achromatic corrosion products covering roughly 80% of the aluminum and galvanized-steel surface. Ablation experiments replicated over five random seeds show that neither contribution claimed from a single run survives replication: synthetic augmentation changes macro-F1 by +0.050 (p = 0.46) under the adopted checkpoint-selection rule and by −0.059 (p = 0.43) under the rule used in the original experiments, and the monotonicity-consistency loss by −0.011 (p = 0.87) and +0.001 (p = 0.99) respectively; the previously reported single-run values of 0.208 and 0.494 are draws from opposite tails of the same seed distributions (0.403 ± 0.140 and 0.344 ± 0.073). The one formulation that improves significantly is the continuous one adopted here, which raises Spearman agreement with the independent metric from 0.316 ± 0.150 to 0.698 ± 0.108 (p = 0.005). Measured against controls, a classifier that never sees the image reaches macro-F1 = 0.425 and, after Holm–Bonferroni correction, no deep configuration is distinguishable from it; none exceeds a one-dimensional linear rule on Scorr (0.664); and under leave-one-material-out cross-validation the network does not improve on Scorr used directly as a predictor (ρ = +0.627 against +0.744, paired p = 0.14). Time-resolved energy-dispersive X-ray spectroscopy (EDS) provides a partial chemical consistency check, with welding at ρ = 0.82 (raw p = 0.023), but no material survives Holm correction across the six tested. A U-Net segmentation head supervised only by synthesis-derived masks attains Dice = 0.85 in-domain and collapses to a 0.033 output range on real images, 5% of the HSV metric’s range; the photometric-stability advantage previously claimed for it is an artifact of that collapse and is withdrawn. Kinetics-aware MCDM with propagated uncertainty resolves 9 of 15 pairwise orderings, placing stainless steel above welding at 30 chamber days with probability 1.000 and reversing the static ranking. The pipeline, code and fixed data split are fully reproducible (random seed 42).
Open AccessSynthetic-Data-Augmented Corrosion-Severity Grading of Grounding Connectors: A Colorimetric Benchmark and Kinetics-Aware Ranking
Corrosion-severity grading of grounding-grid connectors from optical images supports proactive power-infrastructure maintenance. Existing approaches rely on single-time-point, manually thresholded hue–saturation–value (HSV) metrics and static multi-criteria decision-making (MCDM) frameworks that cannot capture corrosion dynamics. In this paper we present a pipeline that (1) defines a four-class corrosion grade from an HSV area fraction (Scorr) measured on RGBA optical images, and validates those labels against a baseline-referenced CIEDE2000 metric zero-referenced to each connector’s as-received appearance; (2) generates 240 color-prior-constrained procedural synthetic images from 53 real images across six connector types; (3) fine-tunes a ResNet-18 to estimate corrosion coverage continuously, deriving the reported severity class from that estimate rather than predicting it directly; and (4) fits power-law kinetics C(t) = k·tn to the Scorr time series, propagates bootstrap uncertainty into a Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) framework, and reports kinetics-aware rankings as rank probabilities. The label validation quantifies two limitations of single-threshold HSV grading: a material-color offset that scores an unexposed copper connector at Scorr = 0.442, and insensitivity to achromatic corrosion products covering roughly 80% of the aluminum and galvanized-steel surface. Ablation experiments replicated over five random seeds show that neither contribution claimed from a single run survives replication: synthetic augmentation changes macro-F1 by +0.050 (p = 0.46) under the adopted checkpoint-selection rule and by −0.059 (p = 0.43) under the rule used in the original experiments, and the monotonicity-consistency loss by −0.011 (p = 0.87) and +0.001 (p = 0.99) respectively; the previously reported single-run values of 0.208 and 0.494 are draws from opposite tails of the same seed distributions (0.403 ± 0.140 and 0.344 ± 0.073). The one formulation that improves significantly is the continuous one adopted here, which raises Spearman agreement with the independent metric from 0.316 ± 0.150 to 0.698 ± 0.108 (p = 0.005). Measured against controls, a classifier that never sees the image reaches macro-F1 = 0.425 and, after Holm–Bonferroni correction, no deep configuration is distinguishable from it; none exceeds a one-dimensional linear rule on Scorr (0.664); and under leave-one-material-out cross-validation the network does not improve on Scorr used directly as a predictor (ρ = +0.627 against +0.744, paired p = 0.14). Time-resolved energy-dispersive X-ray spectroscopy (EDS) provides a partial chemical consistency check, with welding at ρ = 0.82 (raw p = 0.023), but no material survives Holm correction across the six tested. A U-Net segmentation head supervised only by synthesis-derived masks attains Dice = 0.85 in-domain and collapses to a 0.033 output range on real images, 5% of the HSV metric’s range; the photometric-stability advantage previously claimed for it is an artifact of that collapse and is withdrawn. Kinetics-aware MCDM with propagated uncertainty resolves 9 of 15 pairwise orderings, placing stainless steel above welding at 30 chamber days with probability 1.000 and reversing the static ranking. The pipeline, code and fixed data split are fully reproducible (random seed 42).
- Open Access
Multi-Criteria Evaluation and Scenario-Driven Selection of Grounding Connectors Across Materials and Joining Processes
Grounding connectors critically influence the safety and long-term reliability of earthing systems through coupled electro-thermal, mechanical, and corrosion behaviors, yet no standardized quantitative framework exists for jointly evaluating these performance dimensions across diverse deployment scenarios. This study introduces a unified multi-criteria evaluation framework applied to six grounding connector configurations spanning four alloy families and three joining technologies. Electro-thermal response was characterized by coupled finite element simulations (0–100 A), mechanical reliability by quasi-static tensile testing (n = 10 per configuration), and corrosion durability by accelerated salt-spray exposure with image-based corroded area fraction quantification. Performance metrics were normalized and aggregated using equal-weight, Analytic Hierarchy Process, and Shannon entropy weighting schemes, with the Technique for Order of Preference by Similarity to Ideal Solution applied for multi-scenario ranking. One-way analysis of variance confirmed statistically significant effects of connector type on tensile performance (F(5, 54) = 3154.90, p < 0.001). The exothermic welded joint achieved the highest mean ultimate tensile load (61.5 ± 1.5 kN), while copper mechanical connectors exhibited the lowest steady-state temperature rise (~2 K above ambient at 100 A). Compression-crimped connectors ranked first under both equal and Analytic Hierarchy Process weighting (closeness coefficients 0.737 and 0.807, respectively), while stainless steel connectors ranked first under corrosion-critical deployment scenarios. Scenario-weighted analyses demonstrate that the optimal material–process combination shifts with environmental severity, current duty, and mechanical demand, providing a reproducible, evidence-based basis for context-dependent connector specification.
Open AccessMulti-Criteria Evaluation and Scenario-Driven Selection of Grounding Connectors Across Materials and Joining Processes
Grounding connectors critically influence the safety and long-term reliability of earthing systems through coupled electro-thermal, mechanical, and corrosion behaviors, yet no standardized quantitative framework exists for jointly evaluating these performance dimensions across diverse deployment scenarios. This study introduces a unified multi-criteria evaluation framework applied to six grounding connector configurations spanning four alloy families and three joining technologies. Electro-thermal response was characterized by coupled finite element simulations (0–100 A), mechanical reliability by quasi-static tensile testing (n = 10 per configuration), and corrosion durability by accelerated salt-spray exposure with image-based corroded area fraction quantification. Performance metrics were normalized and aggregated using equal-weight, Analytic Hierarchy Process, and Shannon entropy weighting schemes, with the Technique for Order of Preference by Similarity to Ideal Solution applied for multi-scenario ranking. One-way analysis of variance confirmed statistically significant effects of connector type on tensile performance (F(5, 54) = 3154.90, p < 0.001). The exothermic welded joint achieved the highest mean ultimate tensile load (61.5 ± 1.5 kN), while copper mechanical connectors exhibited the lowest steady-state temperature rise (~2 K above ambient at 100 A). Compression-crimped connectors ranked first under both equal and Analytic Hierarchy Process weighting (closeness coefficients 0.737 and 0.807, respectively), while stainless steel connectors ranked first under corrosion-critical deployment scenarios. Scenario-weighted analyses demonstrate that the optimal material–process combination shifts with environmental severity, current duty, and mechanical demand, providing a reproducible, evidence-based basis for context-dependent connector specification.
- 14 April 2026
- Published by Springer Nature in Journal of Real-Time Image Processing
- Vol. 23 (2) , 87
AFENet: a real-time power line and transmission tower segmentation network with adaptive feature enhancement
- 14 April 2026
- Published by Springer Nature in Journal of Real-Time Image Processing
- Vol. 23 (2) , 87
AFENet: a real-time power line and transmission tower segmentation network with adaptive feature enhancement
- Open Access
- 1 December 2025
- Published by MDPI AG in Applied Sciences
- Vol. 15 (23) , 12733
Optimization of HOTAS Joystick Design Based on Virtual Simulation
To address the lack of effective ergonomic evaluation methods for HOTAS (Hands-On Throttle And Stick) control sticks in dynamic operation scenarios, this study proposes a virtual hand motion data acquisition system based on Unreal Engine 5, combined with Leap Motion tracking technology. This system captures and records hand movements and interaction processes in real time, providing key quantitative indicators such as movement trajectories, operation durations, and key positioning deviations. Through four-stage HOTAS task experiments, the efficiency and accuracy differences of various operation types are quantitatively analyzed to provide data-driven support for the ergonomic optimization of HOTAS control rods. The experimental results verify the effectiveness of the system for dynamic human factors evaluation, and it is expected to serve as a preliminary virtual evaluation tool for the ergonomic optimization of HOTAS.
- 1 December 2025
- Published by MDPI AG in Applied Sciences
- Vol. 15 (23) , 12733
Open AccessOptimization of HOTAS Joystick Design Based on Virtual Simulation
To address the lack of effective ergonomic evaluation methods for HOTAS (Hands-On Throttle And Stick) control sticks in dynamic operation scenarios, this study proposes a virtual hand motion data acquisition system based on Unreal Engine 5, combined with Leap Motion tracking technology. This system captures and records hand movements and interaction processes in real time, providing key quantitative indicators such as movement trajectories, operation durations, and key positioning deviations. Through four-stage HOTAS task experiments, the efficiency and accuracy differences of various operation types are quantitatively analyzed to provide data-driven support for the ergonomic optimization of HOTAS control rods. The experimental results verify the effectiveness of the system for dynamic human factors evaluation, and it is expected to serve as a preliminary virtual evaluation tool for the ergonomic optimization of HOTAS.
- 10 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-17
Combining Unsupervised Domain Adaptation and Semi-Supervised Learning for Power Line and Transmission Tower Segmentation
Fully supervised image segmentation can effectively extract power line (PL) and transmission tower (TT) from aerial images. However, its performance is constrained by the lack of sufficiently detailed and high-confidence annotations. Furthermore, PL is the hard sample due to its slender shape and low proportion of feature information. To address the aforementioned challenges, this work innovatively introduces unsupervised domain adaptation (UDA) and semi-supervised learning (SSL) into the PL and TT segmentation task, and designs a new framework named UDASSL-Seg. Specifically, UDA is employed for pretraining, enabling the segmentation network to learn generic features and knowledge of hard sample. Subsequently, SSL is employed for fine-tuning, enabling the segmentation network to acquire generalization capabilities on the target dataset. Additionally, in order to further augment the segmentation network’s performance, the new designed dynamic co-perturbation consistency (DCPC) was proposed to extend the perturbation space by combining multiple image-level and dynamic feature-level perturbations. Extensive experiments were conducted on both self-built and public datasets. The results demonstrate the superiority of the proposed UDASSL-Seg over several state-of-the-art semi-supervised segmentation methods.
- 10 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-17
Combining Unsupervised Domain Adaptation and Semi-Supervised Learning for Power Line and Transmission Tower Segmentation
Fully supervised image segmentation can effectively extract power line (PL) and transmission tower (TT) from aerial images. However, its performance is constrained by the lack of sufficiently detailed and high-confidence annotations. Furthermore, PL is the hard sample due to its slender shape and low proportion of feature information. To address the aforementioned challenges, this work innovatively introduces unsupervised domain adaptation (UDA) and semi-supervised learning (SSL) into the PL and TT segmentation task, and designs a new framework named UDASSL-Seg. Specifically, UDA is employed for pretraining, enabling the segmentation network to learn generic features and knowledge of hard sample. Subsequently, SSL is employed for fine-tuning, enabling the segmentation network to acquire generalization capabilities on the target dataset. Additionally, in order to further augment the segmentation network’s performance, the new designed dynamic co-perturbation consistency (DCPC) was proposed to extend the perturbation space by combining multiple image-level and dynamic feature-level perturbations. Extensive experiments were conducted on both self-built and public datasets. The results demonstrate the superiority of the proposed UDASSL-Seg over several state-of-the-art semi-supervised segmentation methods.
- 12 March 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-10
Power Line Detection From Bottom Up Based on the Characteristic Keypoint
In low-flying unmanned aerial vehicle (UAV) missions, power line detection and localization are crucial to ensure flight safety. Traditional methods make it difficult to detect the global shape and spatial location of power lines due to the elongated nature of power lines, interference from complex backgrounds, and extremely low signal-to-noise ratio due to the high resolution of aerial-measured power line images. To address this problem, this paper proposes a bottom-up power line detection method based on feature keypoints. We utilize the HRNet backbone network to capture multi-scale features and design a power line feature aggregation module (LFA) to enhance the correlation between keypoints. Directional correlation is introduced to enhance the accuracy of keypoint connections further and improve the algorithm’s anti-interference. With the keypoint heatmap prediction and geometric decoder, we can accurately recognize the keypoints of power lines and connect them to a complete power line structure. The experimental results show that the method is superior to the traditional segmentation model in terms of accuracy, and the method achieves a detection accuracy of 0.8321, performs well in high aspect ratio and low signal-to-noise ratio environments, and has a broad application prospect and important research value.
- 12 March 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-10
Power Line Detection From Bottom Up Based on the Characteristic Keypoint
In low-flying unmanned aerial vehicle (UAV) missions, power line detection and localization are crucial to ensure flight safety. Traditional methods make it difficult to detect the global shape and spatial location of power lines due to the elongated nature of power lines, interference from complex backgrounds, and extremely low signal-to-noise ratio due to the high resolution of aerial-measured power line images. To address this problem, this paper proposes a bottom-up power line detection method based on feature keypoints. We utilize the HRNet backbone network to capture multi-scale features and design a power line feature aggregation module (LFA) to enhance the correlation between keypoints. Directional correlation is introduced to enhance the accuracy of keypoint connections further and improve the algorithm’s anti-interference. With the keypoint heatmap prediction and geometric decoder, we can accurately recognize the keypoints of power lines and connect them to a complete power line structure. The experimental results show that the method is superior to the traditional segmentation model in terms of accuracy, and the method achieves a detection accuracy of 0.8321, performs well in high aspect ratio and low signal-to-noise ratio environments, and has a broad application prospect and important research value.
- Open Access
- 11 March 2025
- Published by MDPI AG in Applied Sciences
- Vol. 15 (6) , 3019
Optimizing Joystick Button Layout for Males: An Incremental Approach Based on Gripping Operational Features
Button layout plays a crucial role in user–device interaction. An impractical layout can hinder user efficiency and satisfaction. Current joystick button designs lack systematic references, relying on common body measurements and subjective judgments, which makes it difficult to cope with diverse modes of operation. This paper introduces a basic grip state based on adult male hand size, defining three operational areas. We conducted experiments involving 15 participants at three hand length percentiles (10, 50, 90), revealing positive correlations of operation range with body dimensions. Further discussion analyzes the dynamic operating range and specific features in different areas. These findings inform an incremental design principle for joystick layouts, offering precise guidance for future designs.
- 11 March 2025
- Published by MDPI AG in Applied Sciences
- Vol. 15 (6) , 3019
Open AccessOptimizing Joystick Button Layout for Males: An Incremental Approach Based on Gripping Operational Features
Button layout plays a crucial role in user–device interaction. An impractical layout can hinder user efficiency and satisfaction. Current joystick button designs lack systematic references, relying on common body measurements and subjective judgments, which makes it difficult to cope with diverse modes of operation. This paper introduces a basic grip state based on adult male hand size, defining three operational areas. We conducted experiments involving 15 participants at three hand length percentiles (10, 50, 90), revealing positive correlations of operation range with body dimensions. Further discussion analyzes the dynamic operating range and specific features in different areas. These findings inform an incremental design principle for joystick layouts, offering precise guidance for future designs.
- Open Access
A Two-Stage Corrosion Defect Detection Method for Substation Equipment Based on Object Detection and Semantic Segmentation
Corrosion defects will increase the risk of power equipment failure, which will directly affect the stable operation of power systems. Although existing methods can detect the corrosion of equipment, these methods are often poor in real-time. This study presents a two-stage detection approach that combines YOLOv8 and DDRNet to achieve real-time and precise corrosion area localization. In the first stage, the YOLOv8 network is used to identify and locate substation equipment, and the detected ROI areas are passed to the DDRNet network in the second stage for semantic segmentation. To enhance the performance of both YOLOv8 and DDRNet, a multi-head attention block is integrated into their algorithms. Additionally, to address the challenge posed by the scarcity of corrosion defect samples, this study augmented the dataset using the cut-copy-paste method. Experimental results indicate that the improved YOLOv8 and DDRNet, incorporating the multi-head attention block, boost the mAP and mIoU by 5.8 and 9.7, respectively, when compared to the original method on our self-built dataset. These findings also validate the effectiveness of our data augmentation technique in enhancing the model’s detection accuracy for corrosion categories. Ultimately, the effectiveness of the proposed two-stage detection method in the real-time detection of substation equipment corrosion defects is verified, and it is 48.7% faster than the one-stage method.
Open AccessA Two-Stage Corrosion Defect Detection Method for Substation Equipment Based on Object Detection and Semantic Segmentation
Corrosion defects will increase the risk of power equipment failure, which will directly affect the stable operation of power systems. Although existing methods can detect the corrosion of equipment, these methods are often poor in real-time. This study presents a two-stage detection approach that combines YOLOv8 and DDRNet to achieve real-time and precise corrosion area localization. In the first stage, the YOLOv8 network is used to identify and locate substation equipment, and the detected ROI areas are passed to the DDRNet network in the second stage for semantic segmentation. To enhance the performance of both YOLOv8 and DDRNet, a multi-head attention block is integrated into their algorithms. Additionally, to address the challenge posed by the scarcity of corrosion defect samples, this study augmented the dataset using the cut-copy-paste method. Experimental results indicate that the improved YOLOv8 and DDRNet, incorporating the multi-head attention block, boost the mAP and mIoU by 5.8 and 9.7, respectively, when compared to the original method on our self-built dataset. These findings also validate the effectiveness of our data augmentation technique in enhancing the model’s detection accuracy for corrosion categories. Ultimately, the effectiveness of the proposed two-stage detection method in the real-time detection of substation equipment corrosion defects is verified, and it is 48.7% faster than the one-stage method.