基于时频图像特征提取的状态识别方法研究与应用

Time-frequency image feature extraction for machine condition classification and its application

介绍一种基于时频图像进行设备状态识别的新方法。论述了对振动信号应用Hilbert谱构建二维时频图像,研究时频图像三维重心与信息熵的特征提取方法构造特征向量,采用支持向量机进行设备的状态识别。并以滚动轴承不同状态的识别为例,验证方法的有效性。为提高滚动轴承振动信号特征提取的可靠性,采用多循环平均方法减小循环波动性影响。研究表明此方法能够提高设备故障诊断与状态识别的准确性。

Here,a new machine condition pattern recognition method was introduced.Hilbert spectrum was used to construct time-frequency image for vibration signal analysis.Time-frequency image gravity centre and information entropy were investigated to construct the feature vector.Support vector machine was used as a classification tool for pattern recognition.Rolling bearing pattern recognition was as an example to verify effectiveness of this method in the research.Multi-cycle synchronous average method was used to reduce cyclostationary effect between different cycles to improve the recognition accuracy.The results showed that this method can raise the precision of machine fault diagnosis and pattern recognition.