Quantitative and redundant multi-variable fault diagnosis in induction motors Article uri icon

abstract

  • In this work, a quantitative multi-variable fault diagnosis algorithm is presented for three-phase induction motors based entirely on electrical measurementssupply voltages and line currents. Redundancy is added by employing two residual functions, which are constructed by the total supply power and Park%27s vector magnitude with the aim of visualizing the fault components in frequency sub-bands. A detailed analysis of the resulting frequency content in these residuals is carried out by assuming that the faults are affecting the line currents by a magnitude modulation effect in steady state. In addition, a complete analysis of the residuals robustness against stator asymmetries due to the manufacturing process of the induction motor is described. Next, using a multi-resolution wavelet decomposition of the residuals, the energy of these sub-bands is accurately computed and compared with a baseline condition to detect quantitatively a fault scenario. In this way, variable speed and load torque operating conditions could be monitored for the induction motor. Finally, simulation and experimental evaluations are carried out for an asymmetric rotor fault to show the applicability of the diagnosis scheme. Copyright © Taylor %26 Francis Group.
  • In this work, a quantitative multi-variable fault diagnosis algorithm is presented for three-phase induction motors based entirely on electrical measurementssupply voltages and line currents. Redundancy is added by employing two residual functions, which are constructed by the total supply power and Park's vector magnitude with the aim of visualizing the fault components in frequency sub-bands. A detailed analysis of the resulting frequency content in these residuals is carried out by assuming that the faults are affecting the line currents by a magnitude modulation effect in steady state. In addition, a complete analysis of the residuals robustness against stator asymmetries due to the manufacturing process of the induction motor is described. Next, using a multi-resolution wavelet decomposition of the residuals, the energy of these sub-bands is accurately computed and compared with a baseline condition to detect quantitatively a fault scenario. In this way, variable speed and load torque operating conditions could be monitored for the induction motor. Finally, simulation and experimental evaluations are carried out for an asymmetric rotor fault to show the applicability of the diagnosis scheme. Copyright © Taylor %26 Francis Group.

publication date

  • 2011-01-01