Fault diagnosis and failure prognosis for engineering systems: A global perspective

Rotor-Bar Breakage Mechanism and Prognosis in an Induction Motor

“…1) The first one compares the amplitude of the low sideband harmonic, LSH, to the amplitude of the fundamental component from the current spectrum (1). In order to increase its accuracy, a Blackman window was used to avoid leakage effects and the Teager-Kaiser algorithm, having a minimal computational cost, was utilized to enhance the automatic peak detection [24] in an interval determined by the encoder reading.…”

“…Following the system architecture depicted in [1,6], a Particle Filtering approach is implemented as diagnosis module, which permits the combination of both Boolean and continuous variables, the use of non-linear models (7) and an easy linking to prognosis routines, among other advantages.…”

“…13, right side) has been implemented according to [1,6] as well. Two assumptions are made for its implementation:…”

“…In order to achieve this objective, a complete Condition-Based Maintenance and Prognostics and Health Management (CBM/PHM) system must developed, requiring a better comprehension of the physics behind the failure and their evolution. Incipient fault detection and failure models are critical stages in this methodology since, for the first case, prognostics rely on early detection to carry out a timely and trustworthy prediction, therefore the fault's detection threshold should be reduced as far as it is practical; and secondly, the development of physics-based modeling of components allows a better understanding of the influence of external factors, the failure mechanisms and its propagation times [1].…”

“…Particle Filtering has been developed as a powerful method with low computational cost [3], combining a model and data driven approaches [4][5][6], where model deviations are corrected by new measurements using Bayesian theory, having already been applied to predict the Remaining Useful Life (RUL) of equipment [1,6] The aim of this paper consist of, following the framework outlined in [1], introducing a full CBM/PHM procedure for diagnosing and estimating the RUL of a rotor bar in an induction motor, based on the results of a fatigue test intended to develop in the most natural way a bar breakage. The architecture proposed here ( Fig.…”