摘要
针对风机工作中由于高海拔地理位置、恶劣天气等因素的影响,致使风机叶片出现裂纹、沙眼等缺陷故障,提出基于YOLOv3算法的风机叶片故障检测模型。将风机叶片缺陷区域具有YOLO格式的数据集划分为训练集与测试集,输入YOLOv3模型进行实验,结果表明:YOLOv3模型与YOLOv2模型相比,精度提升3.7%,达到了90.6%;召回率提升3.2%,达到了90.5%;精度平均值提升4.8%,达到了76.2%。
Considering the fact that high-altitude location,bad weather and other factors will trouble fan blades and cause defects like cracks and trachoma,a YOLOv3 algorithm-based fault detection model was proposed which has the data set with YOLO format in fan blade's defect region divided into a training set and a test set,and the has the YOLOv3 model input to the experiment.The results show that,compared with the YOLOv2 model,the accuracy of the YOLOv3 model can be improved by 3.7%and reach 90.6%;the recall rate increased by 3.2%to 90.5%and the average accuracy improved by 4.8%and reach 76.2%.
作者
朱玉廷
汪怡然
马锦雄
谢鹏
陆鹏
汤占军
山子岐
ZHU Yu-ting;WANG Yi-ran;MA Jin-xiong;XIE Peng;LU Peng;TANG Zhan-jun;SHAN Zi-qi(Faculty of Information Engineering and Automation,Kunming University of Science and Technology;Yunnan Longyuan New Energy Co.,Ltd.)
出处
《化工自动化及仪表》
CAS
2024年第3期487-494,共8页
Control and Instruments in Chemical Industry
基金
国家能源集团科技创新项目(批准号:CSIEKJ230700101)资助的课题。