油库离心泵振动信号分离与故障识别研究

Research on Vibration Signal Separation and Fault Diagnosis of Centrifugal Pump in Oil Depot

摘要: 本文利用多通道振动信号采集仪，采集得到某油库各种类型振动信号，通过信号的数字化音频测试技术进行振动信号的识别与特征信号的截取，并采用信号的基元分段法实现振动信号合理、科学地分段处理。采用信号的模式滤波法对振动信号进行时频子波分解，噪声过滤，不同水平子波的聚类，分类子波信号重构与识别，以及信号的归类汇总处理。研究表明，可以将实测振动信号分成基底和泵体振动、电机轴承振动、流固耦合振动、泵零部件松动、背景干扰等13个大类的分类信号；不同工况下泵体振动分类信号的时域参数统计结果具有良好的空间分布特征，可以利用这些信息实现油库工况识别、离心泵状态检测与故障诊断处理。本文研究为振动测试技术在油库安全检测与油库机械设备振动故障诊断中的应用，创造良好的条件。

Abstract: This paper acquires various types of vibration signal of a depot site with using a multi-channel vibration signal acquisition system. These vibration signals have been recognized by use of digital audio signal interception test technology, and then these signals have been segmented reasonably by the Basic Operation Unit Method. On this basis, Pattern Filtering Method is adopted to decompose vibration signal under different working conditions into the time-frequency wavelet while noise is also filtered at the same time. All wavelets are classified into characteristic clustering under different correlation levels through Pattern Filter Analysis. Signal reconstruction and signal recognition of different clustering wavelet are accomplished as well. Finally, vibration signals have been split into different functional and characteristic signal after the source of all clustering wavelet are determined reasonably. It’s shown that vibration signals of an oil depot can be divided into the basement vibration, motor bearing vibration, vibration of centrifugal pump, loosening vibration of pump parts, background interference and other 13 categories. These time domain parameters of every signal separation have good spatial distribution pattern under different conditions. Therefore, using these feature information can realize oil depot working state identification, centrifugal pump’s state detection and fault diagnosis. This study provides good conditions for the application of vibration test in oil depot safety testing and oil mechanical equipment vibration fault diagnosis.