Motor Fault Detection Based On Rough Sets And Support Vector Machine

The squirrel-cage asynchronous motor is used widely in the national production, because of its merits such as simple structure, low price, etc. Along with the improvement of the stator windings of the motor manufacturing process failure rate declined, squirrel cage motor rotor manufacturing process for decades no major change in rotor failure will become one of the main reasons leading to asynchronous motor failure. Asynchronous motor rotor fault, when the rotor a slight fracture of the initial motor performance will be affected but the motor can start and rotation, this time in production generally are not aware of and care. Fault motor is not repair and adjustment, electrical failure further deterioration eventually leading to the motor can not start rotating the production equipment downtime. In this paper, on the basis of inheriting a large number of research results, a rough set of data pre-processing and application support vector machine-based induction motor rotor broken bar fault diagnosis method.First analysis of the asynchronous motor principle and common faults, including the motor stator fault, rotor failure, bearing failure, and the characteristic signal generated when the failure mechanism and failure of the induction motor; Secondly, the principle of rough setintroduction, describes the definition of rough set knowledge, knowledge of the indiscernibility relation in rough set attribute reduction and asynchronous motor fault condition, the discrete signal data, rough set of data reduction; again design the hardware system, including sensors, the choice of the data acquisition card, and the position of the sensor configuration, describes the schematic diagram of the data acquisition card and input and output terminals. A brief analysis of the difference between the differential input and single-ended input and features. Finally, use rough sets to preprocess data and support vector machine combined with asynchronous motor rotor bar breaking fault and the rotor bar with end ring breakage, rotor eccentricity sweep The inner surface of motor, rotor imbalance fault diagnosis. And neural networks, support vector machine, the two diagnostic methods were compared, the results showed that:This method not only greatly improve the speed of diagnosis, and also increased the accuracy rate of induction motor rotor fault diagnosis.