基于微流体与图像识别技术的润滑油磨粒分析方法

发布时间：2018-05-31 10:44

本文选题：在线磨粒监测 + 微流控技术　；参考：《大连海事大学》2017年硕士论文

【摘要】：磨粒检测是滑油检测项目中重要一项,磨粒携带机器设备的磨损信息,能够反映机器的磨损状态。对磨粒进行分析可以判断机器的运转状态,进而为机器的视情维修提供依据。本文在研究了大量现有磨粒在线监测技术的基础上,总结各种磨粒在线监测技术的不足,结合微流控技术的优势,创新性的提出了基于微流体与图像识别技术的润滑油中磨粒分析方法。(1)将微流体技术与铁谱技术相结合的方法实现磨粒检测。用微流控芯片取代传统铁谱片并显微成像,实现润滑油的检测。此种方法集成微流控技术和铁谱技术的优点,既分析便捷、试剂消耗少,又能实时观测磨粒的形态、大小和其它特征,从而实现迅速确定设备磨损类型。(2)设计了基于显微图像技术的实验系统,可以在线监测润滑油污染度和磨粒大小及形状。围绕磨粒图像定性分析所必须解决的图像处理、特征提取和磨粒识别三个基本问题展开了一系列的工作。首先在图像处理上,应用维纳滤波恢复模糊图像,采用差值图像粗分割和Otsu算法相结合的方法实现磨粒图像分割;其次在特征提取上,提取磨粒的6个形态学特征参数包括:面积、周长、长轴、短长轴之比、圆形度和紧凑度。除此之外,本文运用分行维数描述磨粒表面形貌特征,并创新性的采用HSV格式的颜色特征参数色度均值和亮度均值判断磨粒的氧化情况,较以往RGB模式更加简单,使得对磨粒分析较以往更加多样化;最后在磨粒识别上,运用灰色关联分析理论,实现磨损颗粒的自动识别。应用本实验系统不仅可以定量的检测润滑油的污染程度,也可以判断磨粒的类型,集两种检测于一体,其准确率达到92%。
[Abstract]:Abrasive particle detection is an important item in oil testing. Abrasive particles carry the wear information of the machine equipment and can reflect the wear state of the machine. The analysis of the abrasive particles can judge the running state of the machine and provide the basis for the maintenance of the machine. On the basis of studying a large number of existing on-line monitoring techniques of wear particles, this paper summarizes the shortcomings of various on-line monitoring techniques of wear particles, and combines the advantages of microfluidic technology. An innovative method of wear particle analysis in lubricating oil based on micro-fluid and image recognition technology is proposed. The method of combining micro-fluid technology with ferrography to detect wear particles is proposed. The microfluidic chip was used to replace the traditional ferrography chip and the microimaging was carried out to detect the lubricating oil. This method integrates the advantages of microfluidic technology and ferrography technology. It is easy to analyze, consume less reagent, and can observe the morphology, size and other characteristics of abrasive particles in real time. An experimental system based on microscopic image technology is designed to monitor the contamination degree of lubricating oil and the size and shape of abrasive particles on line. A series of work has been carried out on the three basic problems of image processing, feature extraction and wear particle recognition, which must be solved in the qualitative analysis of wear particle images. Firstly, in image processing, Wiener filter is used to restore fuzzy image, then the coarse segmentation of difference image is combined with Otsu algorithm to realize the segmentation of abrasive image. The six morphological characteristics of the extracted particles include: area, circumference, long axis, ratio of short long axis, roundness and compactness. In addition, in this paper, the line dimension is used to describe the surface morphology of the wear particles, and the color feature parameters of HSV format are innovatively used to judge the oxidation of the wear particles. Compared with the previous RGB mode, it is much simpler to judge the oxidation of the particles. The analysis of wear particles is more diversified than before. Finally, the grey correlation analysis theory is used to realize the automatic recognition of wear particles. This experiment system can not only quantitatively detect the contamination degree of lubricating oil, but also judge the type of abrasive particles.
【学位授予单位】：大连海事大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U672;TP391.41

【相似文献】