| Resumen |
This article describes a novel computational model for electric fault diagnostic in induction motors. The essential concept is that a minimum electric fault, like inter-turn short circuit, produces a slight variation that can be identified in current and rotor speed signals. This model uses motor data catalogue to calculate constant parameters that are handled in an original mathematical algorithm that employs varying parameters as function of motor slip. The model performs electric fault simulation and with them, are obtained operation characteristics that build relative and absolute patterns for normal and fault operation. These patterns train a neural network that accomplish the diagnostic in its phase implementation. |
