Evaluation and Classification of Double Bar Breakages Through Three-Axes Vibration Sensor in Induction Motors

The relative positions of broken bars can potentially change the current signature content, filter some of the components and hence lead to misleading in diagnostic process. Especially, if the positions of broken bars are in half pole pitch distance, the characteristic sidebands harmonics get lower which leads to decrease diagnostics ability of stator current based analysis. This paper proposes to analyze double bar breakages faults by monitoring the three-axes vibration (
-x
,
-y
,
and -z
) signals through an accelerometer in detail. The characteristics fault signatures are presented in axial
-x
, radial
-y
and gravity
-z
axis vibrations and a neural network-based classifier (Multi-Layer Perceptron) is utilized to classify the type of double bar breakages. The findings are verified through the experiments. It is shown that some of characteristics fault signatures such as
$2sf_{s}$
,
$3sf_{s}$
and
$2f_{r}$
±
$2f_{s}$
at corresponding vibration spectra can provide more reliable result to detect and classify the broken bar fault in induction motors (IMs).