Relational Map Between Vibration and Corresponding Current Signatures in Induction Motors

Vibration signatures are routinely used for condition monitoring of rotary induction machines. The sources of vibrations can directly be traced to the mechanical state of the machine. Thus, providing a direct link to its health. Nonetheless, collecting vibration data may not always be feasible due to accessibility constraints or because of the presence of other sources of vibrations and noise in the environment. Therefore, the last couple of decades have seen a development of electric current-based condition monitoring for such machines. Therefore, in this article an attempt has been made to illustrate how the two modalities relate to one another. A relational map from electrical (current and voltage waveforms) domain to mechanical (vibration) domain is proposed. This relational map can then be exploited to setup a transfer learning-based process to utilize electrical measurements (current and voltage) and use them to classify faults, usually best detected in mechanical (vibration) domain.