一种融合在线铁谱图像特征信息的磨损状态诊断方法

本资料包含pdf文件1个，下载需要1积分

Using on-line ferrograph image to measure the machine wear status is the core and bottleneck of online visual ferrograph technology. Aiming at on-line oil monitoring for abrasive image information obtained, we designed an integration of different characterizations of on-line iron wear of the image feature information analysis. The method first grayed the original image, the histogram threshold binary, gaussian salt and pepper to enhance, sharp- en the template, several times expansion and corrosion, the boundary tracking, automatic threshold segmentation to obtain more accurate quantitative statistics abrasive image information to measure the concentration of wear. Secondly, re-integration of energy, entropy, moment of inertia, local smooth texture and other characteristics of the wear of the image analysis, diagnosis, evaluation. Finally, the use of RBF neural network technology into the es- tablished classification of the iron spectrum abrasive for automatic identification, experimentation verificated the feasibility of this method.采用在线铁谱图像表征机器磨损状态是铁谱诊断技术的核心和瓶颈.针对在线油液监测获取的磨粒图像信息,设计了一种融合在线铁谱图像特征信息的磨损状态诊断方法,首先通过对原始图像进行灰度化、直方图阈值二值化、高斯与椒盐增强、模板锐化、多次膨胀与腐蚀、边界跟踪、自适应阈值分割处理,获得较为准确的图像磨粒量化统计质量分数信息来量度相对磨损浓度;再融合能量、熵、惯性矩、局部平稳性等图像纹理特征对磨损状态进行分析、诊断、评价;最后采用RBF神经网络技术对铁谱磁性磨粒进行自动识别.实验验证了该方法的创造性和可行性.