Skip to Main Content
During the last decades, non-invasive techniques have been proposed to carry out the fault diagnosis in electric machines. When these techniques are employed in the fault detection of the induction machine's rotors, they exhibit a strong dependence on factors such as the motor load inertia or the opposed torque. In order to develop automatic diagnostic systems or the diagnosis assistance over the rotor state, such as expert systems or knowledge based systems, it is necessary to have available further information to weigh up the influence of these factors. This work presents a study done, based on induction motor's mathematical models, about the incidence of the motor inertia and the opposed torque in some non-invasive fault detection technique employed more frequently. The model takes into account in an independent way each of the rotor bars and then allows to represent different faulty situations. These techniques with non-invasive features do not require sensors directly over the motor and thus allow a diagnosis even on-line when the machine is running. Particularly the study of the following techniques was approached: power spectral analysis, torque spectral analysis, stator current spectral analysis and Park's current vector behaviour. Therefore the diagnostic may be based in the measurement of motor's external variables such as applied voltages and stator currents.