Chair: Prof. Hyunseok Oh (Gwangju Institute of Science and Technology)
A Time-varying Energy Residual (TVER) Method for Fault Detection of a Planetary Gear Under Variable Speed Conditions Mr. Jungho PARK, Moussa HAMADACHE, Jong M. HA, Yunhan KIM and Byeng D. YOUN (Seoul National University)
In recent decades, many methods have been developed to detect faults in the gear system. In particular, techniques for fault detection of planetary gears have gained considerable attention because planetary gears are often exposed to harsh operating conditions, such as heavy loads or high speeds. However, many previous studies are limited to detection of the gear faults under a constant speed condition. In real-world settings, geared systems often operate under a variable speed condition. This study thus proposes a time-varying energy residual (TVER) method to detect faults in a planetary gear under a variable speed condition. In the proposed method, the wavelet transform is exploited to extract energy-related features that can express time-varying and faulty behaviors of the gear vibration signals in the time-frequency domain. Then, the Gaussian process is used with the wavelet coefficients to calculate energy residuals that represent the fault severities of transient signals. The effectiveness of the proposed method was demonstrated through a case study of simulated vibration signals of the planetary gear. From the case study, it can be concluded that the proposed TVER method can successfully differentiate a faulty gear from a normal one in a variable speed condition.
Feature extraction for gear diagnostics based on EEMD in different crack size Mr. Seok Ju HAM, Sung-Ho PARK and Joo-Ho CHOI (koera areo space university, Korea Areospace university)
In these days, diagnostics techniques enabling the condition based maintenance are being paid great attention in many industry fields in order to achieve increased reliability of the system as well as the reduction of operating cost. The techniques are particularly useful to the system that costs a tremendous amount for the maintenance or leads to the catastrophic results when failure occurs. In such systems, the gearbox is usually employed to deliver the power under extreme loading conditions as are found in the helicopter, wind turbines and many others. The gearbox is expensive to maintain and replace when failure occurs. There are two types of gear fault. One is the spall that chips off the surface of the teeth and the other is the crack that is formed at the root of tooth due to the repeated bending stress. Cracks are more critical in the sense that it grows suddenly to the tooth breakage, resulting in the whole system loss. In the previous study, we have conducted fault classification of spall and crack using the Ensemble Empirical Mode Decomposition (EEMD) technique based on the transmission error (TE) signals, in which the faults are imbedded to the gear, and the difference of the faults is identified with the aid of finite element analysis (FEA) of the faulted gears. In this study, further progress is made with the goal to evaluate the severity of the crack faults from the signal. To this end, gears with different crack size are prepared. The FEAs are conducted and compared with the measured signals, from which the critical size is identified that requires maintenance action. Since the measured crack signals include various noise and uncertainties, study on the statistical significance is also made to check whether the signal can be large enough to detect the fault. Once successful, the technique can be applied to estimate not only the size of the crack fault but also its severity against the critical level from the TEs of the gears in operation.
Evaluation of a Knock Sensor for Gearbox Diagnosis Mr. Keon KIM, Jong M. HA, Jungho PARK and Byeng D. YOUN (Seoul National University)
A gearbox is one of the critical components in rotating machinery. Timely prediction of gearbox faults become of great importance to minimizing unscheduled machine downtime. Most of gearbox diagnosis studies are focused on the development of gearbox diagnosis algorithms using costly vibration sensors. However, vibration sensor cost matters in some applications, thus pushing to the use of a low-cost accelerometer, such as a knock sensor. This study develops a sensor evaluation process for the purpose of diagnosis. This study uses a planetary gearbox with a knock sensor, known to be cheap and good for high frequency applications (i.e., diesel engines). First, gearbox feature engineering is thoroughly studied through time domain and frequency domain analyses. This study uses the features that are most sensitive to gearbox faults, such as pitting and surface damages. Second, some sensor evaluation metrics (i.e., signal-to-noise ratio (SNR)) are overviewed for the purpose of diagnosis. The sensor evaluation study suggests a frequency range of the knock sensor that can be reliably used for fault diagnosis. Two case studies are presented to demonstrate the effectiveness of the proposed sensor evaluation process and metric: 1) one-stage planetary gearbox and 2) a swing reduction gear (two-stage planetary gearbox) in an excavator. It is concluded that a knock sensor can be used for the fault diagnosis of a gearbox.
Tooth-wise Fault Identification for a Planetary Gearbox Based on a Health Data Map Mr. Jong M. HA, Jungho PARK and Byeng D. YOUN (Seoul National University, Seoul National University / OnePredict Inc.)
Vibration-based fault diagnostics of a planetary gearbox is known to be challenging due to revolving planet gears. Signal transfer path between the planet gear of interest and the vibration sensor is periodically varying, and thus the vibration signal is modulated. To detect the faults in the planet gear, vibration signals are extracted using a window function when the signal transfer path is minimized. The extracted signals are transferred to the tooth domain of the planet gear of interest for further analysis. However, due to various uncertainties such as manufacturing and assembly tolerances that can significantly affect the vibration characteristics, vibration signals could have an unexpected modulation characteristics regardless of the signal transfer path. It means that the features from the faulty tooth of the gear can be discarded by the abuse of the window function during the extraction procedure. To overcome these challenges, this paper proposes an original idea of a tooth-wise fault identification for a planetary gearbox based on a health data map. In doing so, vibration signals are processed with time synchronous averaging (TSA) and auto-regressive minimum entropy deconvolution (AR-MED) filter that don’t require the use of the window function. Vibration signals processed in time-domain must be aligned in the domains of a pair of gear teeth (i.e. ring-planet gear teeth pairs). Two-dimensional health data map can sketch the health data corresponding to every pairs of gear teeth to isolate the location of the faulty gear tooth. For demonstration of the proposed method, this paper presents case studies using an analytical model and a gearbox testbed.