基于图像识别技术的旋转机械振动故障诊断方法  

The Diagnosis Method of Vibration Faults of Rotating Machinery Based on Image Recognition

作　　者：陈非[1] 施青 黄来[1] 韩彦广[1] 程贵兵[1] 焦庆丰[1] 

机构地区：[1]国网湖南省电力公司电力科学研究院,长沙410007 [2]国电克拉玛依发电有限公司,克拉玛依834003

出　　处：《汽轮机技术》2014年第5期364-366,共3页Turbine Technology

摘　　要：目前振动信号的分析主要是针对特定测点在某一瞬间采集的一段振动波形,提取其中的特征量来进行故障诊断,这种基于状态信息的诊断方法对故障类型的辨别能力有限。通过构造基于归一化幅值谱的三维矩阵来刻画旋转机械振动故障的过程变化规律,并引入图像识别技术,在此基础上建立了一种基于归一化幅值谱图识别的旋转机械振动故障诊断方法。并通过该方法对试验台获取的振动故障信号进行分析,计算结果表明该方法是一种有效的故障诊断方法。At present, vibration signal analysis is a foundation of fault diagnosis, while the common used method is to analyze the characteristic quantities extracted from an original vibration waveform of a particular measuring point in a particular moment. It lacks of abilities in fault classification. The process change rule of each vibration fault of rotating machinery is reflected by constructing a three-dimensional matrix based on standardized amplitude spectrum in this paper, at the same time,the image recognition method is introduced. On this basis, a fault diagnosis method of rotating machinery based on image recognition of standardized amplitude spectrum is proposed. The method has been used to analyze vibration signals and diagnose faults of a rotor test platform, the numeric result indicates that it is an effective fault diagnosis method.

关 键 词：旋转机械 故障诊断 过程 图像识别 

分 类 号：TK26[动力工程及工程热物理—动力机械及工程] O235[理学—运筹学与控制论]