A Graph-based Method to Extract Broken-Rotor-Bar Fault Signature in Varying Speed Operation

Motor current signature analysis (MCSA) has been a mature technique for fault detection in line-fed induction motors for decades. Since nowadays most induction motors are driven by inverters, it is challenging to detect faults in inverter-fed induction motors due to the operation of varying speed and varying load. In this paper, we study broken-bar fault detection in inverter-fed induction motors under varying speed conditions. By representing the stator current using a complex space vector, we define a new fault signature of continuous varying frequency. To effectively extract the newly defined broken-bar fault signature, we propose a graph-based method in which we solve an optimization problem of graph model with constraints imposing smoothness and sparsity of the spectrum. Simulation results demonstrate the effectiveness of the proposed method.