Aiming at the problems of scene complexity and high-precision requirements existing in the defect detection of wind turbine blades, this paper proposes an improved model based on YOLOv11n. Firstly, the lightweight module C3k2_DWR_DRB is designed. Through dynamic multi-scale feature fusion and structural reparameterization technology, the feature expression ability is enhanced while reducing the computational complexity. Secondly, the FDPN_TADDH architecture is introduced, combined with the focused diffusion pyramid network and the task-aligned dynamic detection head, to achieve the precise location and classification of multi-scale defects. The Inner-MPDIoU loss function is further proposed. By optimizing the bounding box regression in the core area, the robustness of the model in locating complex morphological defects is enhanced. Experiments show that the improved model achieves 87.0% and 61.4% respectively in the mAP50 and MAP50-95 metrics, increasing by 2.2% and 7.9% compared with the benchmark model YOLOv11n, and the parameter quantity is only 2.5M. The results show that the proposed method has significant advantages in detection accuracy and lightweight design, and can provide reliable technical support for the intelligent operation and maintenance of wind turbine blades.