Development of Novel Acoustic Emission Based Methodology and Tools for Bearing Fault Diagnostics

2015-10-21T00:00:00Z (GMT) by Brandon E. Van Hecke
Acoustic emission (AE) has proven to be an effective nondestructive technique to investigate the behavior of material under mechanical stress. Compared with vibration techniques, AE offers several advantages. For example, AE techniques are capable of incipient fault detection. Additionally, it is sensitive to fault location, allowing its use for fault location detection. When applied to rolling element bearings, it has been shown that AE techniques can detect faults earlier than other technologies. However, there are a number of challenges in the implementation of AE techniques. Namely, in comparison with vibration sensors, AE sensors require a much higher sampling rate. Additionally, it requires significant storage and imposes a computational burden when the volume of data is large. Lastly, the non-stationary behaviors of AE signals make traditional frequency analysis methods ineffective. In this research, novel AE based methodology and tools are developed that combine a heterodyned based frequency reduction technique, time synchronous resampling, and spectral averaging for bearing fault diagnostics. AE signals from seeded fault bearings are acquired simultaneously with a tachometer signal using a frequency reduction technique on a bearing test rig. Using the crossing times of the tachometer, the signal is time synchronously resampled and spectrally averaged. Finally, after computation of several condition indicators (CIs), bearing fault diagnosis can be achieved. For experimental purposes, a data acquisition system was developed to enable the testing of the proposed methodology. For validation, steel type 6205-2RS bearings were seeded with inner race, outer race, cage, and ball faults for data collection. Then, both AE and accelerometer data was acquired and processed to validate the fault diagnosis capability of the proposed methodology. The results indicate that the proposed signal processing technique efficiently and effectively detects all of the various bearing fault types at both high and low shaft speed ranges. The methodology has also been extended and validated for bearing fault diagnosis and planet gear carrier fatigue crack detection on a UH-60A helicopter using accelerometer data. The outcome of this research is effective and efficient methodology and tools that extract bearing fault features for bearing fault diagnosis with validation using the AE signals of seeded fault bearings.