Search

An interturn short-circuit in the stator windings can lead to the breakdown of electrical machines. In the case of induction machines, several fault detection methods and faulted models have been developed in the recent decades. These models differ mainly in how the leakage inductances of the faulted winding are modeled. This work provides a generalized model for interturn short-circuit faults, using different assumptions for the leakage inductances. The model is validated with experimental results for an exhaustive set of fault parameters, and the influence of leakage inductances is analyzed. Moreover, the model is used to analyze the drawbacks of the negative-sequence fundamental current as a traditional fault signature. A high-frequency injection method for converter-fed machines is presented to overcome these limits. The proposed fault signature is the negative-sequence current at the injection frequency and it is evaluated experimentally at different operating conditions. The fault severity and its location are proved to be related to the proposed fault signature.

Online methods for stator insulation monitoring have been investigated with increasing interest in the last 15 years. These methods rely on the high-frequency behavior of the electric machine. In converter-fed machines, the high-frequency behavior is visible through the current ringings occurring during the switching commutations of the power electronics. As the lifetime of electric machines can last several decades, a faster way to evaluate insulation monitoring methods is through accelerated aging tests. This work aims to perform an accelerated aging procedure on an induction motor to test an online insulation monitoring method. The motor has a randomly-wound stator winding and it is aged by means of thermal cycles without the need for an oven. A custom converter is employed during the accelerated aging procedure to apply the thermal cycles and to record the current ringings. The proposed insulation monitoring method is based on a metric that quantifies the current ringing variations in a simple manner, following the quasi-linear variation of the parasitic capacitances during the motor aging.

Condition monitoring of the insulation health of electric machines is usually performed with offline tests. In recent years, online condition monitoring methods for the groundwall insulation have been introduced. The majority of these works are based on leakage current measurements, requiring additional equipment compared to the standard one in converters. The condition monitoring method in this work instead relies on the high-frequency ringing of stator currents occurring after a converter switching transition. This work evaluates the method experimentally in a randomly-wound stator winding of an induction machine prototype equipped with taps. The insulation condition change is forced by inserting external capacitors. Through the analysis of different metrics for the health monitoring and a detailed discussion of the hardware and software requirements for an electric drive, a viable way forward for the practical implementation of the approach is proposed.