New condition monitoring procedure for lifting cranes using motor current signal analysis

Vibration analysis does not cover all defects able to affect a rotating machine, including those that induce torque or instantaneous rotation speed fluctuations. Indeed the vibrational energy contributions of this type of defects such as a natural and gradual degradation of the gears are often weak. Therefore the appearance of these defects is very difficult to detect from the vibrational energy indicators commonly used in machine condition monitoring. In these conditions, it is often late when the symptoms of failure are reflected in the vibration signal. Several signal processing techniques have emerged to improve the vibration diagnostic result by proposing the use of angular resampling techniques to overcome the difficulties of analyzing nonstationary signals as, for example, velocity fluctuations that can hide the appearance of defects. Other researchers suggest the direct measurement of torque to make the diagnosis, however this technique is intrusive and too expensive. In the recent years there were large interests in diagnosis techniques using motor current signal analysis since these signals are known to be most representative of the torque, easily acquired, not expensive and especially not intrusive. This paper deals with a new diagnostic method based on current signal analysis by associating a simple demodulation technique with an angular resampling technique generally used for vibration signals. Finally the performance of the new method is illustrated through experimental data treating gears wear detection in lifting cranes.