Lane Maria Rabelo Baccarini

Incipient fault detection in induction machine stator-winding using a fuzzy-Bayesian change point detection approach

In this paper the incipient fault detection problem in induction machine stator-winding is considered. The problem is solved using a new technique of change point detection in time series, based on a three-step formulation. The technique can detect up to two change points in the time series. The first step consists of a Kohonen neural network classification algorithm that defines the model to be used, one change point or two change points. The second step consists of a fuzzy clustering to transform the initial data, with arbitrary distribution, into a new one that can be approximated by a beta distribution. The fuzzy cluster centers are determined by using the Kohonen neural network classification algorithm used in the first step. The last step consists in using the Metropolis-Hastings algorithm for performing the change point detection in the transformed time series generated by the second step with that known distribution. The incipient faults are detected as long as they characterize change points in such transformed time series. The main contribution of the proposed approach in this paper, related to previous one in the Literature, is to detect up to two change points in the time series considered, besides the enhanced resilience of the new fault detection procedure against false alarms, combined with a good sensitivity that allows the detection of rather small fault signals. Simulation results are presented to illustrate the proposed methodology.

Sliding Mode Observer for Rotor Faults Diagnosis

Although induction motors are traditional thought to be reliable and robust, the possibility of faults is unavoidable once the machines can be exposed to different hostile environments, misoperations, and manufacturing defects. Therefore, motor monitoring incipient fault detection and diagnosis are important topics. This paper presents a method for on-line induction motor monitoring with the purpose of detecting and locating a single rotor broken bar. The method avoids any frequency analysis and observes instead the machine state with the help of the two models. The torque difference between the two models indicates a fault. The technique utilizes input signals from standard transducers. An experimental setup has been constructed to implement the new technique in on-line model

The design of multiple linear regression models using a genetic algorithm to diagnose initial short-circuit faults in 3-phase induction motors

Abstract Induction motors are robust machines that are often exposed to a variety of environmental and operating conditions that can result in a number of failures during their use. One such fault is a short-circuit that starts in a few turns, and quickly extends to other winding sections. Early detection and diagnosis of this type of failure is very important and can prevent the complete motor loss. In this work, a multiple linear regression modelling technique is used in synergy with the genetic algorithm optimization and the analysis of variance methods to obtain models to classify the motor operating in normal and in short-circuit conditions. The proposed method is suitable for application in real industrial plants due to three important features: (i) it uses RMS values of voltages and currents, (ii) only simulation data is required to obtain the MLR classification model and (iii) incipient faults can be identified with high accuracy. Experimental tests carried out over a wide range of machine operation conditions demonstrates the simplicity and effectiveness of the new diagnosis method.

Fault detection and diagnosis in duals converters applied in DC drives

Summary form only given. For duals converters of AC or DC drives, faults related with them are important in relation with the available time of equipments. Besides, the speed sensors used as tachogenerator and pulse generator are important factors. This work presents a fault diagnostic and detection system for commutation, short circuit and circuit open faults. Speed sensor faults can also be detected with an automatic feedback reconfiguration. The fault diagnostic and detection system is based in the fuzzy set theory to generate inputs to the neural network responsible for fault classification, and the state observer for sensor speed fault detection. Results obtained from digital simulations conclude that the system is a simple and efficient means for the detection of the proposed faults. Other advantage is that the control system reconfiguration based on the state observers permits the drive system to operate during speed sensor faults. So, this proposed fault detection system is ideal for practical implementation.