| Detections of faults in squirrel-cage rotors is very difficult to motor manufacturers. Statistics have indicated, 10% fault in induction motors are causes by the rotor faults. When the rotor is broken, motor quality will be affect seriously. So, it is necessary to analyse rotor quality.The traditional detection method is by means of the test of stator current and torque when motor is running. Therefore rotor faults can be detected form analysis these data. But its test cycle is long, also the influence for motor performance factor is very much. So this method doesn't suit for the production line to analyse rotor quality. In order to detect rotor faults more quickly, a new method is developed, which is based on induction magnetic field. These studies are performed using nonlinear time series analysis and the chaotic data analysis to generate fault case data. Inspired by data mining, this method employs time-delayed embedding and identifies the largest Lyapunov exponent and the correlation dimension in the resulting phase spaces.This method is more effective than using the tradition method to detect rotor faults. Moreover, rotor can be detected before it is assembled to motor. We can detect interrupted bars, air enclosures, porosity and wrong skew for rotors. It provided one simple fast examination method for the rotor mass production. Hence, this method is more advantageous to the increase production efficiency and reduce the test cost than to use the torque or stator current detection. |