Abstract:In the realm of special equipment, significant advancements have been achieved in fault detection. Nonetheless, faults originating in the equipment manifest with diverse morphological characteristics and varying scales. Certain faults necessitate the extrapolation from global information owing to their occurrence in localized areas. Simultaneously, the intricacies of the inspection area's background easily interfere with the intelligent detection processes. Hence, a refined YOLOv8 algorithm leveraging the Swin Transformer is proposed, tailored for detecting faults in special equipment. The Swin Transformer serves as the foundational network of the YOLOv8 framework, amplifying its capability to concentrate on comprehensive features during the feature extraction, crucial for fault analysis. A multi-head self-attention mechanism regulated by a sliding window is utilized to expand the observation window's scope. Moreover, an asymptotic feature pyramid network is introduced to augment spatial feature extraction for smaller targets. Within this network architecture, adjacent low-level features are merged, while high-level features are gradually integrated into the fusion process. This prevents loss or degradation of feature information during transmission and interaction, enabling accurate localization of smaller targets. Drawing from wheel–rail faults of lifting equipment as an illustration, the proposed method is employed to diagnose an expanded fault dataset generated through transfer learning. Experimental findings substantiate that the proposed method in adeptly addressing numerous challenges encountered in the intelligent fault detection of special equipment. Moreover, it outperforms mainstream target detection models, achieving real-time detection capabilities.

What problem does this paper attempt to address?

The paper primarily focuses on addressing the issue of wheel-rail fault detection in special equipment (such as cranes). Specifically, the research proposes improved methods to tackle the following challenges: 1. **Diverse Fault Forms and Scales**: Faults in special equipment exhibit diverse morphological characteristics and different scales. 2. **Need for Local Information**: Certain types of faults require extracting local details from global information for analysis. 3. **Background Complexity**: The complex background of the detection area can easily interfere with the intelligent detection process. To address these challenges, the paper proposes an improved algorithm based on YOLOv8 and Swin Transformer. The main features of this algorithm include: - **Swin Transformer as the Base Network**: Introducing Swin Transformer into the YOLOv8 framework to enhance the ability to focus on comprehensive features during the feature extraction process. - **Multi-Head Self-Attention Mechanism**: Utilizing a sliding window-based multi-head self-attention mechanism to expand the observation window range. - **Asymptotic Feature Pyramid Network (AFPN)**: Introducing AFPN to enhance the spatial feature extraction of small targets. This network structure allows the fusion of adjacent low-level features and gradually incorporates high-level features into the fusion process, thereby avoiding the loss or degradation of feature information during transmission and interaction. Experimental validation shows that the proposed improved method can effectively address multiple challenges encountered in intelligent fault detection of special equipment and surpasses mainstream object detection models in terms of real-time detection capability. Additionally, the method leverages the transfer learning strategy of the wheel-rail fault dataset to improve the model's accuracy and robustness, especially in cases with limited sample sizes. In summary, this research aims to achieve effective and accurate detection of wheel-rail faults in special equipment by combining the advantages of YOLOv8 and Swin Transformer, as well as introducing techniques such as AFPN.

An Improved Method for Detecting Crane Wheel–Rail Faults Based on YOLOv8 and the Swin Transformer

Detection Algorithm for Bearing Roller End Surface Defects Based on Improved YOLOv5n and Image Fusion

TBi-YOLOv5: A Surface Defect Detection Model for Crane Wire with Bottleneck Transformer and Small Target Detection Layer

A Fault Detection Method for Transmission Line Components Based on Synthetic Dataset and Improved YOLOv5

Swin-Transformer -YOLOv5 for lightweight hot-rolled steel strips surface defect detection algorithm

An Improved YOLOv8 Algorithm for Rail Surface Defect Detection

STAE‐YOLO: Intelligent detection algorithm for risk management of construction machinery intrusion on transmission lines based on visual perception

Improved YOLOv8 for B-scan Image Flaw Detection of the Heavy-Haul Railway

Rail Wheel Tread Defect Detection Using Improved YOLOv3

Improved swin transformer-based defect detection method for transmission line patrol inspection images

A rail fastener defect detection algorithm based on improved YOLOv5

Swin-RGC: Swin-Transformer with Recursive Gated Convolution for substation equipment non-rigid defect detection

Track Fastener Defect Detection Model Based on Improved YOLOv5s

Lightweight Algorithm for Rail Fastener Status Detection Based on YOLOv8n

Improved YOLOv8-Based Target Precision Detection Algorithm for Train Wheel Tread Defects

Improved YOLOv7-based steel surface defect detection algorithm

Transformer Discharge Carbon-Trace Detection Based on Improved MSRCR Image-Enhancement Algorithm and YOLOv8 Model

A Defect Detection Method for Grading Rings of Transmission Lines Based on Improved YOLOv8

RAM-YOLOv8: A multi-fault detection method for photovoltaic module

CRGF-YOLO: An Optimized Multi-Scale Feature Fusion Model Based on YOLOv5 for Detection of Steel Surface Defects

High Speed Railway Fastener Defect Detection by Using Improved YoLoX-Nano Model