Informative frequency band identification method using bi-frequency map clustering for fault detection in rotating machines

Jacek Wodecki,Konstantinos Gryllias,Agnieszka Wylomanska,Piotr Kruczek,Radoslaw Zimroz,Justyna Hebda-Sobkowicz

doi:10.21595/vp.2018.20210

Abstract

In presented work the problem of local damage detection in rolling element bearings is addressed. Usually such issues require the usage of the techniques of decomposition, separation etc. In such real industrial cases main difficulty lies in relatively low signal-to-noise ratio as well as unpredictable distribution of damage-related information in the frequency domain, hence the typical methods cannot be used. In this paper such industrial scenario is addressed and a simple yet effective approach to underlying component extraction will be discussed. Proposed method analyzes Cyclic Spectral Coherence map as starting data representation, and Expectation-Maximization is used as analytical tool to determine the informative frequency band (IFB) for impulsive component localization in the carrier frequency spectrum. Finally, based on identified IFB, the bandpass filter is constructed to extract the impulsive component from the input signal.

Highlights

MethodologyThe proposed procedure consists of three main steps. Firstly, the bi-frequency representation of the vibration signal is calculated
Proposed method analyzes Cyclic Spectral Coherence map as starting data representation, and Expectation-Maximization is used as analytical tool to determine the informative frequency band (IFB) for impulsive component localization in the carrier frequency spectrum
The boundaries of the cluster of the CSC map, that can be interpreted as containing the informative frequency band, are used to construct a bandpass filter for impulsive component extraction

Summary

Methodology

The proposed procedure consists of three main steps. Firstly, the bi-frequency representation of the vibration signal is calculated. The boundaries of the cluster of the CSC map, that can be interpreted as containing the informative frequency band, are used to construct a bandpass filter for impulsive component extraction. All these techniques are well-known, their combination is proposed as a successful approach in condition monitoring, focusing on cases, where an impulsive damage-related component investigation is needed but on the other hand a meaningful determination of the IFB is difficult using simpler methods. The definitions and the main properties of the proposed methodology are briefly introduced. Their fusion is illustrated and described afterwards

Cyclic spectral coherence

Expectation-Maximization clustering algorithm

Application to industrial data

Conclusions