Modelling of Broken Rotor Bars and Eccentricity Faults in Squirrel Cage Induction Machines

Induction machines are used in a wide range of industrial applications due to their simplicity, ruggedness, and low price. Despite their robustness, they eventually fail at one point due to a variety of failure mechanisms. Many faults, such as broken rotor bars and eccentricity faults, produce specific fault frequencies in the motor current spectrum, which allows for fault detection. Model-based fault detection methods compare the measured quantities during operation of the machine to a machine model given the same inputs. Depending on the fault type, model parameters can be adapted to match the faulty machine behavior. This allows to estimate fault severity from the model parameters.Classical models based on the Clarke transformation are not able to replicate the asymmetrical fault states investigated in this work. Therefore, a model based on modified winding functions and multiple coupled circuits is used. It allows to flexibly emulate many different fault states and matches the behavior of the machine over a wide operating range, even in faulty state. The developed model is evaluated using steady-state measurements on an inverter-fed 5.5 kW induction machine.