Research on Wind Turbine Power Curve Anomaly Detection Technology Based on YOLOv8

This article aims to provide a method based on image recognition that can quickly locate power curve abnormalities of wind turbines. Power curves are often used to evaluate the power generation performance of wind turbines. Wind turbines with abnormal performance can be detected by observing their power curve images. This project uses the existing labelimg software and yolov8 architecture to classify, label and train a model for a large number of power curve images with abnormal characteristics. This model is then used in batch power curve characteristic screening. In application scenarios that require large-scale screening, compared with the traditional use of human eyes and brains to reason about power curve pictures, the application of this model can quickly locate abnormalities and abnormal types of power curves in batches.