Study On Fault Features Extraction For Helicopter Rolling Bearings Based On DTCWT And ICA

Transmission system is one of the three key components in helicopter. Rolling bearings are kernels of helicoprter transmission system and rolling bearing faults were often found in helicopter transmission system. So it is very important to detect such faults early to ensure the safety and reliability of the helicopter.Traditional helicopter rolling bearing fault diagnosis methods include Temperature Method, Oil-like Analysis Method, Direct disassembled and detection Method and so on. But those methods can not meet the need of Condition-based Mainteance. Vibration Method is a new fault diagnosis method developed in recent years, it has better detection results for rolling bearings incipient fault and requires more simple equipments than other methods, so it has became a focus research topic. One of Vibration Method's key techniques is feature extraction of fault signals. Fault signals of rolling bearings in helicopter are often complex due to their complex structures and serious operating circumstances. Traditional methods are less suitable to extract the effective fault features readily. So it is valuable to study how to extract rolling bearing fault features effectively.This paper presents a new fault features extraction method for helicopter rolling bearings based on DTCWT and ICA.The main work of this paper is as follows.1. The main research on fault features extraction of helicopter rolling bearings is analyzed and the disadvantages of the traditional methods are summarized.2. The main failure modes and virbration mechanism of helicopter rolling bearings are analysed.3. Basic theories about DTCWT (Dual-Tree Complex Wavelet Transform, DTCWT) and ICA (Independent Component Analysis, ICA) are introduced. Comparative analysis shows that DTCWT has less spectrum aliasing and better translational invariant than wavelet transform method in the 1-D signal processing. The advantages of ICA in signal de-noising and mixed signals separation are analysed.4. The method based on DTCWT and ICA is deeply studied to extract rolling bearing fault features. Compared with Fourier transform method, Wavelet transform method, Laplacian Pyramid transform method and Resonant Demodulation method theoretically and experimentaly, the method presented in this paper is more effective and efficient.