Search

Over the years induction motors have established an uncanny knack for providing a plethora of utilities in the industry, where the fault monitoring and detection has become necessary. Several techniques could be applied for the monitoring and identification of broken rotor bars when the motor is fed by a variable speed drive (VSD). Nevertheless, many of these methodologies detect this fault and other failures in the steady state condition, but this monitoring grow into more complicated analysis during the startup transient condition owing to the large number of harmonics, which the VSD insert to the current signal. The novelty of the proposed methodology is the application of the reassignment during the startup transient and the steady state conditions to identify one broken rotor bar in the induction motor. The proposed methodology is experimented with both, real and synthetic signals. The problems that Short Time Fourier Transform (STFT), shows for the identification of broken rotor bars are exhibited. The proposed methodology includes an automatic diagnosis (K-means algorithm), where the signal energy is used. The results show that the Reassigned Short Time Fourier Transform (RSTFT) technique and K-means methods are appropriated for the effective monitoring and diagnosis of one broken rotor bar in the induction motor during the startup and steady state conditions of operation.

Sensorless speed estimation has been extensively studied for its use in control schemes. Nevertheless, it is also a key step when applying Motor Current Signature Analysis to induction motor diagnosis: accurate speed estimation is vital to locate fault harmonics, and prevent false positives and false negatives, as shown at the beginning of the paper through a real industrial case. Unfortunately, existing sensorless speed estimation techniques either do not provide enough precision for this purpose or have limited applicability. Currently, this is preventing Industry 4.0 from having a precise and automatic system to monitor the motor condition. Despite its importance, there is no research published reviewing this topic. To fill this gap, this paper investigates, from both theoretical background and an industrial application perspective, the reasons behind these problems. Therefore, the families of sensorless speed estimation techniques, mainly conceived for sensorless control, are here reviewed and thoroughly analyzed from the perspective of their use for diagnosis. Moreover, the algorithms implemented in the two leading commercial diagnostic devices are analyzed using real examples from a database of industrial measurements belonging to 79 induction motors. The analysis and discussion through the paper are synthesized to summarize the lacks and weaknesses of the industry application of these methods, which helps to highlight the open problems, challenges and research prospects, showing the direction in which research efforts have to be made to solve this important problem.

The study of power quality (PQ) has gained relevance over the years due to the increase in non-linear loads connected to the grid. Therefore, it is important to study the propagation of power quality disturbances (PQDs) to determine the propagation points in the grid, and their source of generation. Some papers in the state of the art perform the analysis of punctual measurements of a limited number of PQDs, some of them using high-cost commercial equipment. The proposed method is based upon a developed proprietary system, composed of a data logger FPGA with GPS, that allows the performance of synchronized measurements merged with the full parameterized PQD model, allowing the detection and tracking of disturbances propagating through the grid using wavelet transform (WT), fast Fourier transform (FFT), Hilbert-Huang transform (HHT), genetic algorithms (GAs), and particle swarm optimization (PSO). Measurements have been performed in an industrial installation, detecting the propagation of three PQDs: impulsive transients propagated at two locations in the grid, voltage fluctuation, and harmonic content propagated to all the locations. The results obtained show that the low-cost system and the developed methodology allow the detection of several PQDs, and track their propagation within a grid with 100% accuracy.

This paper investigates the current monitoring for effective fault diagnosis in induction motor (IM) by using random forest (RF) algorithms. A rotor bar breakage of IM does not derive in a catastrophic fault but its timely detection can avoid catastrophic consequences in the stator or prevent malfunctioning of those applications in which this sort of fault is the primary concern. Current-based fault signatures depend enormously on the IM power source and in the load connected to the motor. Hence, homogeneous sets of current signals were acquired through multiple experiments at particular loading torques and IM feedings from an experimental test bench in which incipient rotor severities were considered. Understanding the importance of each fault signature in relation to its diagnosis performance is an interesting matter. To this end, we propose a hybrid approach based on Simulated Annealing algorithm to conduct a global search over the computed feature set for feature selection purposes, which reduce the computational requirements of the diagnosis tool. Then, a novel Oblique RF classifier is used to build multivariate trees, which explicitly learn optimal split directions at internal nodes through penalized Ridge regression. This algorithm has been compared with other state-of-the-art classifiers through careful evaluation of performance measures not encountered in this field., (Copyright © 2018 ISA. Published by Elsevier Ltd. All rights reserved.)