在线铁谱图像分析中基于蚁群算法改进Otsu的设计与应用  

Design and application of improved Otsu based on ant colony algorithm on on-line ferrograph image analysis

作　　者：陶辉 陈闽杰 贺石中 冯伟 

机构地区：[1]广州机械科学研究院有限公司设备状态检测研究所,广东广州510700

出　　处：《电子设计工程》2014年第10期60-63,67,共5页Electronic Design Engineering

基　　金：广东省教育部产学研项目(2011B090400488);国机集团发展基金(17130001)

摘　　要：在线铁谱图像获取机器磨损状态信息是铁谱诊断技术的核心和瓶颈。针对在线铁谱磨粒图像的Kirsch边缘检测特征不明显和Otsu(最大类间方差法)获取最佳阈值的局限性及耗时等问题,设计了一种基于蚁群算法改进Otsu方法完成图像分割,并结合Kirsch边缘检测来提取磨粒图像信息的新方法。首先通过Kirsch算子检测出图像边缘,然后运用基于蚁群算法改进Otsu方法求取最佳阈值并进行二值化处理,最后采用灰度堆栈空间实现磨粒自动定位。通过现场对三峡电厂5号水轮发电机组2012年油液进行试验和数据分析、及近一年的机组开机老化运行,得出所设计的算法能够有效提取磨粒图像信息,同时节省运算时间,对水轮机组故障预测、诊断起到了良好的实际作用。Using on-line ferrograph image to obtain the machine wear status is the core and bottlenecks of on-line visual ferrograph technology. Aim to solve shortcomings of the Kirseh edge features of the gray image is not obvious and the Otsu method is time-consuming while it is used to access to the best threshold automatic. This paper presents an application of an improved Otsu method in the Kirsch edge detection. Firstly, it uses the algorithm of Kirsch to detect the edge of the image. Then it calculates the best threshold through the combination of Ant Colony Algorithm and the Otsu method and execute binary processing. At last the abrasive automatic positioning by using gray level stack space. Through experiment and data analysis through the turbine No. 5 turbine for the Three Gorges power plant oil field in 2012, the designed algorithm can effectively extract the wear particle image information, and save the operation time, have a good adaptability to the fault diagnosis of hydro turbine generating unit.

关 键 词：在线铁谱 Kirsch边缘检测 OTSU 蚁群算法 

分 类 号：TK478.9[动力工程及工程热物理—动力机械及工程]