摘要
针对目前诊断系统之间不能通用导致的诊断系统重复开发以及各种振动数据采集仪器、在线式和便携式故障诊断系统都能提供频谱图的现状,分析了各种频谱图的图像特点。提出了利用图像识别进行旋转机械故障特征提取的方法。结合Multi-agent技术开发了基于频谱图像识别的Multi-agent旋转机械故障诊断系统,解决了诊断软件的通用性和适应性之间的矛盾。应用表明,该系统是实用的、有效的。
Faced with the status quo that there exists repetitive development of fault diagnosis systems because of their incompatibility and that vibration data sampling instruments of various types and online or portable fault diagnosis systems can produce spectral images, the paper analyzes the characteristics of spectral images of various kinds and put forward a fault feature extraction method for rotational machinery based on image recognition. In combination with multi-agent technology, it developed the rotational machinery fault diagnosis system based on spectral image recognition, thus solving the contradiction between fault diagnosis software incompabibility and adaptability. An application instance indicates that the fault diagnosis system is effective.
出处
《机械科学与技术》
CSCD
北大核心
2007年第3期283-286,共4页
Mechanical Science and Technology for Aerospace Engineering
关键词
旋转机械
频谱图像
特征提取
图像识别
rotational machinery
spectral image
feature extraction
image recognition
