<p>Wind turbine blades are the core components of wind turbines. Due to the harsh environment in which they are located and long-term high-speed rotation, wind turbine blades are prone to various types of damage. However, traditional wind turbine blade damage detection algorithms have certain limitations, such as low detection accuracy for small damage and too many model parameters, which makes deployment difficult. To solve these problems, this study proposed an improved wind turbine blade damage detection algorithm RIG-YOLOv8n (RFAConv-C2f_IDC-GSA-YOLOv8n) based on the YOLOv8n architecture. The algorithm replaces the original C2f module with the C2f_IDC (C2f with inception depthwise convolution) module to enhance the receptive field and reduce computational overhead, and introduces the RFAConv (receptive-field attention convolution) module to improve sensitivity to small damage through feature rearrangement and weighting mechanism. At the same time, the GSA (group shuffle attention) attention mechanism is integrated to optimize the fusion capability of local and global features, thereby improving target positioning and classification accuracy. Experimental results show that RIG-YOLOv8n improves detection accuracy while significantly reducing model parameters and computational load. To assess deployment feasibility, we tested the model on a local server with an NVIDIA RTX 4090 GPU. RIG-YOLOv8n achieved the fastest inference time, highest FPS, and smallest model size among all compared models, demonstrating strong real-time performance and deployment efficiency.</p>