一种EEMD-HHT与时频重排结合的转速曲线估计新方法

New method for speed curve estimation by using the EEMD-HHT combined with time-frequency reassignment

HHT(Hilbert-Huang Transform)和传统时频分析存在模态混叠、时频聚焦性差等问题,在估计转速曲线时误差较大。因此,提出将EEMD-HHT与时频重排算法结合的转速曲线估计新方法。采用EEMD(Ensemble Empirical Mode Decompositon)方法获取IMFs;找出参考轴所对应的IMF,并做Hilbert变换;用时频重排算法处理该IMF两端各一定长度的数据,对得到的重排矩阵进行限制条件的瘠线搜索,并将获得的瞬时频率值代替对应时间点内Hilbert变换得到的瞬时频率值;对"拼接"得到的整个时间段的瞬时频率值进行最小二乘拟合,得到光滑的转速曲线。将该方法应用于转子不平衡故障的阶比分析中,对实际测量信号进行了分析,并与HHT估计结果作比较,验证了其优越性和有效性。

The HHT and traditional time-frequency analysis methods have usually the disadvantages of mode mixing, low time-frequency concentration and so on, which makes the speed curve estimation error larger. So a new method for estimating the speed curve was proposed by using the EEMD-HHT combined with a time-frequency reassignment algorithm. First, the EEMD method was used to obtain IMFs. The IMF corresponding to the referentiral axis was Looked for, on which the Hilbert transform was carried out. Next, the time-frequency reassignment transformation method was used to process the data at each end of this IMF. The ridge of the reassignment matrix was searched with some limiting conditions, and the obtained instantaneous frequencies at both ends were used to replace those values obtained by the Hilbert transform. Then, the combined instantaneous frequencies were fitted to obtain a smooth speed curve. The method was applied to the order analysis of actual rotor unbalance faults. The validity and superiority of the proposed method were verified by comparing its results with the HHT estimation results.