Identificação de falhas em motores de indução trifásicos utilizando algoritmos de otimização e sistemas inteligentes
Year: 2019
Type: dissertation
Abstract: Three-phase induction motors are essential to the production process, as their construction features allow for wide application, including operation in hostile environments. Although robust, its continuous use generates wear that can lead to failures. Thus, researches in the area of identification of failures are carried out constantly aiming to reduce rates of corrective maintenance and unwanted stops, allowing the machine maintenance. For this purpose, the present work proposes the study and development of a strategy that uses optmization algorithm combined with pattern classifiers to identify short-circuit stator failures, broken bars of rotor and bearing wear in three-phase induction motors, using voltage and phase current signals obtained in the time domain, and the speed module. The algorithms Differential Evolution, Particle Swarm Optimization and Simulated Annealing are used to estimate the electrical parameters of the induction motor using the model of the equivalent electrical circuit and the identification of the failure occurs by variation of these parameters with the evolution of each fault. The classification of each type of failure or its presence is made by pattern classifiers Artificial Neural Network, Support Vector Machine and k-Nearest Neighbor. The database used for the work was obtained through laboratory experiments performed with a motor of 1 HP and 2 HP, under conditions of variation of conjugate of line-connection, including unbalance voltages.
Author: Jacqueline Jordan Guedes
Institution: Universidade Tecnológica Federal do Paraná
Language: Portuguese
Subfield: Control and Systems Engineering
Field: Engineering
Domain: Physical Sciences
Keywords: three-phase induction motor, fault diagnosis, broken rotor bars, bearing wear, differential evolution, particle swarm optimization, simulated annealing, equivalent circuit parameter estimation, artificial neural networks, support vector machine, k-nearest neighbors, voltage and current signals, voltage unbalance, predictive maintenance
Open Access status: green
OpenAlex ID: W3116855537
