基于时频等高图的汽轮发电机组振动故障诊断方法研究

状态监测与故障诊断的主要内容包括故障信号的检测、故障诊断、趋势分析和早期故障预测即预测诊断等。传统的振动故障诊断主要是基于频谱分析的方法，小波变换得出的时频等高图是分析故障信号奇异性的有效工具，为汽轮发电机组的故障检测和诊断提供了新思路。在介绍小波变换理论、Morlet小波和时频等高图的基础上，通过对仿真的汽轮发电机组几种典型振动故障，如不对中、油膜涡动和部件松动等信号进行分析，结果表明时频等高图不仅能够直观检测出信号中的微弱奇异成分，而且还可以有效地对故障进行分类诊断、实现早期故障检测和故障变化趋势分析等。最后，采用时频等高图分析了3种实际机组的振动异常信号。

STUDY ON METHODS OF VIBRATION FAULT DIAGNOSIS BASED ON TIME-FREQUENCE CONTOUR MAP FOR TURBINE GENERATOR UNIT

The condition monitoring and fault diagnosis are mainly consist of fault detection, fault diagnosis, trend analysis, and prognosis etc. The traditional vibration diagnostic method was based on spectrum analysis for the turbine generator unit. The time-frequency contour map by wavelet transform has been playing a significant role in the analysis of singularities, which hide in the signal. First of all, in this paper, the theory of Wavelet Transform, Morlet Wavelet, which is one kind of complex wavelet, and the time-frequency contour map were introduced. Then several simulating vibration signals containing typical faults, such as misalignment, oil whirl and loose of the turbine generator, were transformed based on the Morlet Wavelet. The transform results were plotted in time-frequency contour map, which revealed the faults in the signal. The favorable results obtained from the examples show that the time-frequency contour map is an effective method not only in singularity discovering and localization, but also in fault classification diagnosis, early prediction or prognosis, and fault trend analysis etc. Finally, the real vibration signals in power plants are analyzed by the time-frequency contour map.