Broken Rotor Bar Detection in Induction Motors through Information Entropy Analysis on the Start-up Transient and Steady-State Current Signals

Induction motors are key elements for industrial and manufacturing processes. An unexpected fault can provoke severe consequences about reparation costs, product quality or even security. The breaking of a rotor bar may produce secondary faults that will generate catastrophic damages to the motor or surrounding equipment; hence, early fault detection in induction motors has gained significant relevance in last few years. Related techniques in recent literature for broken bar detection employ the combination of distinct signal processing algorithms increasing their computational cost and making difficult their online implementation. Furthermore, previous approaches focus on the start-up transient current analysis since broken rotor bar detection is easier under this regime than under the steady state. In this work, an experimental examination about the feasibility of using information entropy analysis on the electric current supply to an induction motor is performed, during its start-up transient and its steady state. Obtained experimental results exhibit the high effectiveness of information entropy as low computational cost index, for detecting a healthy motor or a motor with one broken bar under both operational regimes, start-up transient and steady state.