2021 : Broken Rotor Bar Detection In Squirrel Cage Induction Motor Using Empirical Mode Decomposition and Artificial Neural Network

Dimas Anton Asfani S.T., M.T.,Ph.D



Published in


External link






Abstract — Induction Motor is one of electric machine type that mostly used in industry because of its reliability and its simplicity in its construction. Any fault on motor can disturb the production process in an industry and cause many disadvantages. The most common fault in induction motor is broken rotor bar. Motor Current Signature Analysis (MCSA) is one of the mostly used methods for fault detection that has the capability to detect common fault on electric machinery especially induction motor. This paper develops an expert system based on MCSA by utilizing Empirical Mode Decomposition (EMD) analysis that combined with Artificial Neural Network (ANN). Through EMD analysis, the stator current is decomposed into its simple form called Intrinsic Mode Function (IMF). The Time Successive Between Zero Crossing (TSZC) on the IMF is classified base on the motor condition. In the proposed method, ANN is designed to be intelligent classifier. The ANN’s input is the parameter of Probability Density Function (PDF) curve of standard deviation from several zero crossings in the IMF. The result shows that the proposed method is successfully identify the fault up to 97.2% efficiency. However, for severity fault or number of broken rotor bar, the proposed method is able to identify 60.9% cases. Generally, the proposed method is provide better performance when motor is running at high load level when current signal is measured.