自主致密LCD和改进的广义局部频率解调方法

自主致密局部特征尺度分解方法通过确定待分解信号的最小信号极值尺度以度量其信号尺度的方式,采用新增伪极值点均匀化信号尺度的方法,可有效抑制局部特征尺度分解的模态混叠。在广义局部频率解调的基础上,基于强局部加权回归和伪端点方法的原理,提出了改进的广义局部频率解调方法。该方法通过分析广义局部频率和广义局部幅值突变情况,在广义局部频率和广义局部幅值中插入伪端点,并采用强局部加权回归对新增伪端点后的广义局部频率和广义局部幅值进行分段平滑处理,可有效改进广义局部频率和广义局部幅值曲线的平滑效果。采用仿真信号将自主致密局部特征尺度分解与局部特征尺度分解、经验模态分解进行对比,同时将改进的广义局部频率解调与希尔伯特变换、广义局部频率解调进行对比,结果证实了自主致密局部特征尺度分解和改进的广义局部频率解调方法的有效性。对转子碰摩故障实验数据的对比分析结果,进一步验证了该方法的有效性和实用性。

The method of autonomous compact local characteristic scale decomposition and improved general local frequency demodulation

In autonomous compact local characteristic-scale decomposition (ACLCD) method, the least extrema scale of the signal to be decomposed is determined to measure the signal scales, and the pseudo-extrema points are added autonomously to homogenize the signal scales, which effectively restrains the mode mixing of local characteristic-scale decomposition (LCD). A novel improved general local frequency demodulation (IGLFD) method is proposed using the theories of robust locally weighted regression (RLWR) method and pseudo-endpoint (PE) method, on the basis of general local frequency demodulation (GLFD). Through analyzing the saltation of general local frequency (GLF) and general local amplitude (GLA) in IGLFD, the method inserts the pseudo-endpoints in GLF and GLA, respectively, and adopts RLWR method to perform the piecewise smooth processing of the GLF and GLA after inserting the pseudo-endpoints, which can effectively improve the smooth effect of GLF and GLA. The simulation signal experiments were used to compare LCLCD with LCD and empirical mode decomposition (EMD), meanwhile IGLFD was compared with hilbert transform (HT) and General local frequency demodulation (GLFD; the results indicate the effectiveness of the proposed ACLCD and IGLFD methods. Finally, the proposed methods were applied to diagnose the rotor with rub-impact fault successfully; the comparison and analysis of the experiment data further verify the effectiveness and practicability of the proposed methods.