A system for incipient fault detection and fault diagnosis based on MCSA

The paper describes a system for automated detection of incipient faults in induction machines. The system has been based on the Motor Current Signature Analysis method (MCSA) and aimed to be applied in a thermal electric power plant in south Brazil. First, the mechanism of fault evolution is introduced and clarified regarding the most common induction motor faults: stator winding short-circuits, broken and cracked rotor bars and eccentricity faults. The influence of the load condition on the fault indicator is discussed based on practical cases, obtained through fault simulations using a prototype. The main theoretical and conceptual aspects of the developed system are presented, including the signal acquisition and conditioning as well the database which stores the motor signals acquired over a time period. Some results from the practical use of the system are shown to illustrate the system capabilities.