Condition monitoring (CM) is the process that assesses the health of equipment/systems at regular intervals or continuously and exposes incipient faults if any. Bearing failure is one of the foremost causes of breakdown in rotating machine, resulting in costly systems downtime. This chapter presents an application of health index (HI) for fault diagnosis of rolling element bearing (REB) which has been successfully used in diverse fields such as image processing, prognostic health management (PHM), and involves integration of mathematical and statistical concepts. There is hardly any effort done in developing HIs using different aspects of wavelet transform (WT) for fault diagnosis of REB. A comparison of the performances of the identified approaches has been made to choose the best one for REB fault diagnosis.
TopIntroduction
Condition monitoring (CM) consists of extraction of information about particular parameters from machines and analysis of data to predict the health of the machines, without disturbing their operation (Jardine et al., 2006). Rolling Element Bearing (REB) are critical components widely used in rotary machines, which will perpetually produce a range of faults due to harsh working environment and complex operating conditions, disturbing the safe and stable operation of the machine. Industries spend millions of dollars for plant maintenance operations and it has been reported that maintenance costs may account for as much as 1/3rd of the manufacturing costs of the product (Abdusslam, 2012). The components of a typical REB are “outer ring, rolling elements, cage and inner ring”. REB defects are classified as localized defects and distributed defects. Localized defects most commonly occur and include cracks, pits and spalls caused by fracture on the rolling surfaces. Distributed defects include surface roughness, waviness, misaligned races and off-sized rolling elements (Lin et. al., 2017, Pirra, 2012, Tandon, & Choudhury, 1999). Different methods are used for revealing and identification of REB faults (Alguindigue et. al., 1999, Tandon & Nakara, 1992). They are wear analysis, temperature, acoustic and vibration measurements. Among these, vibration signature analysis is widely used as reported in (Patil, M. S. et. al., 2008). A major share of the research work on REB fault diagnosis is based on signal processing techniques which primarily include Time, Frequency and Time-Frequency domain techniques. The details and use of time and frequency domain technique can be found in (Abdusslam, 2012, Lin et. al., 2017, Pirra, 2012, Tandon & Nakara 1992, Patil, M. S. et. al., 2008, Tyagi, 2008). Fast Fourier Transform (FFT) is not appropriate for transient vibration signal analysis, as it is unable to reveal transient / non-stationary information contained in it. These transient components contain vital information about bearing defects. Hence, Wavelet Transform (WT) is commonly used to analyze them as reported in (Randall & Anthony, 2011, Smith & Randall, 2015, Castro et. al., 2008, Peng & Chu, 2004) and use of Time Frequency Analysis (TFA) technique can be found in (Tyagi, 2008, Castro et. al., 2008, Smith & Randall, 2015). The WT is a preferred TFA technique, when compared to other techniques, as it provides flexibility in the window and, analyzes high frequency signal with a short duration function waveform (Randall & Anthony, 2011, Smith & Randall, 2015). An extensive review on WT and its applications is available in (Peng & Chu, 2004, Yan et. al., 2014). Based on the signal decomposition paradigms, WT can be classified as “Continuous Wavelet Transform (CWT), Discrete Wavelet Transform (DWT) and Wavelet Packet Transform (WPT)”. Researchers have used them individually or have combined them suitably (for e.g. CWT and WPT) for fault diagnosis (Peng & Chu, 2004, Yan et. al., 2014, Vijay, 2013).