Diagnosis of Mechanical and Electrical System Faults

Many researches have focused on the methods for the diagnosis of mechanical and electrical system failures. However, researchers are still struggling to find a way to test the performance of the mechanical key components such as bearings with different specifications and conduct fault diagnosis in a complete and accurate manner. For this reason, Dr. Yau designed and implemented a bearing test rig to test the four features of a bearing: axial load, radial load, inner ring radial thrust and rotational speed. The test rig can be used to observe the force-bearing status of a bearing and changing the bearings with different specifications can be made quickly. The orthogonal array of the Taguchi method was used to optimize the four features, and an optimization analysis was conducted using scientific methods on the contribution to vibration signals when different forces are applied to the bearing. First, this vibration signal was used to detect the failures of the bearing vibration signal in different failure modes by applying the Chen-Lee chaotic system and the fractal theory. Then, a smart diagnosis of bearing failures was conducted by applying the extension theory. This result was published in the SCI journal IET Science, Measurement & Technology, 2016. Furthermore, Dr. Yau also proposed a detection method that integrates the fractional-order Sprott chaotic synchronization system with the extension theory for the monitoring of ball bearings. Compared with the integral-order Sprott chaotic system, the fractional-order Sprott chaotic system can generate dynamic errors that are easily compatible with the back-end matter-element model of the extension theory, enabling accurate identification of ball bearing signals. The experimental data from the Bearing Data Center of the Case Western Reserve University verified that the method proposed in this study performs better than those adopted in other literatures regarding the accuracy of bearing fault diagnosis, and that the method is able to achieve 100% accuracy when applied to real-time monitoring. This result was published in the SCI journal IEEE Transactions on Industrial Electronics, 2016.