基于多尺度局部对比度的海上风电风机叶片异常点检测  

Detection of Anomaly Points of Offshore Wind Turbine Blades Based on Multi-scale Local Contrast

作　　者：史凯特 李雨欣 任正奎 吕亚利 董秀芬 薛晖晖 王亚辉 宁宁 冉晨 SHI Kaite;LI Yuxin;REN Zhengkui;LV Yali;DONG Xiufen;XUE Huihui;WANG Yahui;NING Ning;RAN Chen(Research Institute of Science and Technology of China Three Gorges Corporation,Beijing 100038,China;School of Intelligent Engineering,Zhengzhou Institute of Aviation Industry Management,Zhengzhou 450015,China;Strategy and Development Research Center of China Three Gorges Corporation,Beijing 100038,China;Lingbao Huaxiang Wind Power Development Co.,Ltd.,Lingbao 472500,China)

机构地区：[1]中国长江三峡集团有限公司科学技术研究院,北京100038 [2]郑州航空工业管理学院智能工程学院,河南郑州450015 [3]中国长江三峡集团有限公司战略与发展研究中心,北京100038 [4]灵宝华祥风电开发有限公司,河南灵宝472500

出　　处：《电工技术》2023年第12期84-87,共4页Electric Engineering

基　　金：中国长江三峡集团有限公司科研项目(编号202003035)。

摘　　要：针对传统方法在海上风电风机叶片砂眼等损伤异常点检测虚警率高的问题,提出了一种基于多尺度局部对比度的风机叶片异常点检测方法。首先通过Facet-Kernel滤波对图像进行预处理,抑制背景杂波,突出异常点;其次通过多尺度局部对比度计算,准确提取异常点位置;最后通过自适应阈值分割提取损伤异常点。实验结果表明,该方法具有较好的检测结果,拥有有效性及鲁棒性。A multi-scale local contrast based anomaly detection method for wind turbine blades is proposed to address the high false alarm rate of traditional methods in detecting abnormal damage points such as sand holes in offshore wind turbine blades.Firstly,the image is preprocessed using Facet Kernel filtering to suppress background clutter and highlight outliers.Secondly,accurate extraction of anomalous point positions is achieved through multi-scale local contrast calculation.Finally,damage outliers are extracted through adaptive threshold segmentation.The experimental results show that this method has good detection results,effectiveness,and robustness.

关 键 词：异常点检测 局部对比度 风机叶片 

分 类 号：TM614[电气工程—电力系统及自动化]