The increasing demand for motors, driven by advancements in industrial automation, underscores the necessity for reliable monitoring systems to ensure optimal operation.

In this context, the study employs the You Only Look Once (YOLO) algorithm for defect detection, facilitating the transfer of imagery data to IoT devices for continuous monitoring.

The research integrates UAV imaging, machine learning algorithms, and IoT technology, enabling real-time inspections with high-resolution cameras.

Conventional manual inspections of large-scale wind turbines are not only inefficient but also pose safety risks, often necessitating turbine shutdowns.

To mitigate these challenges, early detection and repair of surface defects in wind turbine bearings are crucial for minimizing maintenance costs and extending operational life.

Given that wind turbine rolling bearings are vital for the reliability of wind power systems, their failure can lead to significant economic losses.

Accurate diagnosis of bearing faults is essential, yet existing deep learning methods struggle with feature extraction and generalization under complex conditions.

The proposed method demonstrates effectiveness in real-world applications, validated through datasets from Case Western Reserve University and a wind turbine test bench.

Training for the detection model utilized a comprehensive dataset comprising 10,029 annotated images of various defects, including cracks and scratches.

Enhancements to the YOLOv8 model, such as the integration of DSConv for improved geometric feature recognition and an optimized loss function, contribute to its precision in defect detection.

Additionally, the proposed few-shot learning model allows for effective classification even with limited training samples, addressing the challenge of data imbalance.

Overall, the proposed methodologies demonstrate superior early warning capabilities for abnormal conditions, significantly advancing the state of damage detection in wind turbine systems.