Three-phase induction motor fault identification using optimization algorithms and intelligent systems

The present work proposes the study and development of a strategy that uses an optimization algorithm combined with pattern classifiers to identify short-circuit stator failures, broken rotor bars and bearing wear in three-phase induction motors, using voltage, current, and speed signals. The Differential Evolution, Particle Swarm Optimization, and Simulated Annealing algorithms are used to estimate the electrical parameters of the induction motor through the equivalent electrical circuit and the failure identification arises by variation of these parameters with the evolution of each fault. The classification of each type of failure is tested using Artificial Neural Network, Support Vector Machine and k-Nearest Neighbor. The database used for this work was obtained through laboratory experiments performed with 1-HP and 2-HP line-connected motors, under mechanical load variation and unbalanced voltage.