診斷機械故障之結合Hilbert-Huang轉換及階次跟蹤方法

Combine Hilbert-Huang Transform and Order Tracking Methods in Fault Diagnosis of Machinery

沈育霖(Yu-Lin Shen)；林忠賢(Jhong-Sian Lin)；康淵(Yuan Kang)

Hilbert-Huang轉換 ； 經驗模態分解 ； 瞬時頻率 ； 階次跟蹤 ； 故障診斷 ； Hilbert-Huang Transform ； Empirical Mode Decomposition ； Instantaneous Frequency ； Order Tracking ； Fault diagnosis

勞動及職業安全衛生研究季刊

25卷3期（2017 / 09 / 15）

143 - 152

繁體中文

本文針對非穩態運轉的機械探討其故障分析之方法，例如升降速階段因為非等速是一種非平穩過程，無法符合頻譜分析方法對信號平穩性的要求，因此，不能用傅立葉轉換進行分析處理，於是本文使用基於HHT(Hilbert-Huang Transform)的階次跟蹤方法來對非平穩信號進行分析。首先對振動信號進行經驗模態分解得到信號的固有模態函數，再求各個固有模態函數Hilbert 轉換，得到信號的瞬時頻率及瞬時振幅並對應時間即可得HHT(Hilbert-Huang Transform)時頻譜，從時頻譜中取兩個時間點之頻譜，觀察其轉速頻率及共振頻率。將各個瞬時頻率進而求得脈衝信號對應到振動信號之再採樣時間，根據此再採樣時間對原始信號進行等角度重採樣，最後對重採樣信號進行階次跟蹤分析可得到時間階次譜，同樣從時間階次譜中取兩個時間點之階次譜，觀察其階次之幅值變化辨識是否故障。

The paper applies the planet gear deceleration machine for fault diagnosis analysis of study, Because of the gear rises to change down the stage is one kind of unsteady process, is unable to conform to Fourier to transform to the signal stable request, therefore, cannot use conventional the spectral analysis method to carry on the analysis processing, so this use of order tracking based on Hilbert-Huang transform method to analyze unsteady signal. At first carries on the experience modality decomposition to the vibration signal to obtain the signal inherent modality function, Then obtained for each intrinsic mode function of the received signal after Hilbert transform instantaneous frequency and instantaneous amplitude and corresponds to the time then you can get HHT time frequency spectrum then takes spectrum of two time points from the time frequency spectrum, observe the rpm frequency and resonance frequency, and then obtains the signal impulse by each instantaneous frequency to correspond the vibration signal the resampled time, according to this resampled time to the original signal carries on resampled in angle-domain, finally using order tracking to analyze resampling signal. The simulated signal and the example analysis has confirmed this method accuracy and the reliability, then takes order spectrum of two time points from the time order spectrum, observed whether the fault by order amplitude changes.

1. Huang, NE(1998).The empirical mode decomposition and the Hilbert spectrum for nonlinear and non-stationary time series analysis.Proceedings of the Royal Society,454,903-995.
2. Hui, L(2007).Gear Fault Monitoring Based on Order Tracking and Bi-spectrum under Running-Up Condition.Fourth International Conference on Fuzzy Systems and Knowledge Discovery
3. Kang, Y,Wang, CC,Chang, YP,Hsueh, CC,Chang, MC(2006).Certainty Improvement in Diagnosis of Multiple Faults by Using Versatile Membership Functions for Fuzzy Neural Networks.Lecture Notes in Computer Science,3973,370-75.
4. Ruqiang, Y,Gao, RX(2006).Hilbert-Huang Transform-Based Vibration Signal Analysis for Machine Health Monitoring.IEEE Transactions on Instrumentation and Measurement,55,2320-29.
5. Sokolov, RT.,Rogers, JC(1995).Removing Harmonic Signal Nonstationarity by Dynamic Resampling.Proceedings of the IEEE International Symposium on Industrial Electronics,1,303-08.
6. Weidong, J,Shixi, Y,Yongping, C,Gongbiao, Y,Jinsong, H(2008).Detecting a Cracked Rotor with HHT-based Time-Frequency Representation.Proceedings of the IEEE International Conference on Automation and Logistics,Qingdao, China:
7. Yuping, Z(2006).Hilbert-Huang Transform and Marginal Spectrum for Detection of Bearing Localized Defects.The Sixth World Congress on Intelligent Control and Automation
8. Zhidong, Z,Min, P,Yuquan, C(2004).Instantaneous frequency estimate for non-stationary signal.Fifth World Congress on Intelligent Control and Automation