基于改进Yolov4的风电机组叶片缺陷检测算法

Defect Detection Algorithm Technology for the Blade of Wind Turbine Based on Yolov4

新型能源的发展离不开风力发电。随着风电机组性能的不断优化,对叶片运维提出了更高要求。为了提升风电机组的叶片巡检效率,有必要探索自动化、智能化的风电机组叶片巡检技术。无人机巡检技术已在多个领域有所应用,基于挂载高清摄像头的无人机对风电机组叶片所拍摄的近距离图像,通过图像识别实现风电机组叶片缺陷检测。通过研究Yolov4算法在风电机组叶片无人机自动巡检系统中的应用,探索出了提升风电机组叶片缺陷检测精度的新路径。通过深度学习和计算机视觉技术,提高了风电机组叶片检测的实时性、高效性和准确性。通过实验证明,利用数据增强和改进目标检测Yolov4算法,可使风电机组叶片缺陷的检测平均精度(mAP)达83%。

The continuous development of new energy is inseparable from wind power generation.The optimization of wind turbine performance put forward higher requirements of blade operation and maintenance.In order to improve the efficiency of wind turbine blade inspection,it is necessary to explore automatic and intelligent inspection technology for wind turbine blade.Unmanned aerial vehicle(UAV)inspection technology has been applied in many fields.Based on the close range images of wind turbine blades taken by UAV mounted with HD camera,the defect detection of wind turbine blades can be realized through image recognition.By studying the application of Yolov4 algorithm in the UAV automatic inspection system of wind turbine blades,a new path for improving the detection accuracy of wind turbine blade defects is explored.Through deep learning and computer vision technology,the real-time,high efficiency and accuracy of wind turbine blade detection is improved.Experiments show that using Yolov4 algorithm,the average precision of detecting wind turbine blades defect can reach 83%.