Startup-based rotor fault detection in soft-started induction motors for different soft-starter topologies

The use of soft-starters has proliferated in industrial induction motors to damp the negative effects of high-starting currents, among other reasons. Despite the fact that soft-starters reduce the probability of rotor damage, some industrial cases of rotor failures in soft-started motors have been reported. Over recent years, a novel diagnosis trend based on the analysis of the motor startup current is rapidly drawing the attention of the industrial maintenance community due to the important advantages of that method versus other well-known approaches. An interesting variant of that trend relies on the study of some specific wavelet signals resulting from the Discrete Wavelet Transform (DWT) of that current and on the subsequent computation of fault severity indicators. This method was applied with success to the rotor assessment of motors started direct-online and even of certain soft-started induction motors. However, the massive validation of the method in different soft-starter models and with different topologies of their power block was still pending. This issue is solved in the present work which makes use of extensive testing to obtain a huge set of startup signals corresponding to a motor that is started with four different soft-starter variants. The results prove that, despite the identification of the fault components is more difficult when using these drives, it is clearly possible to separate the healthy and faulty condition regardless of the model used.