Abstract: A method and apparatus for detecting rotor faults in an induction motor. Such induction motors have a stator and a rotor comprising a multiplicity of rotor bars. The flux around the rotor bars is detecte at a predefined flux detection point in the motor, generally by using a coil wound around one of the stator teeth. A synchronization signal is generated once per revolution of the rotor when a predefined position on the rotor is closest to a stator reference point, usually the flux detection point. The detected flux signal is filtered to reject signals in a predefined frequency band around the frequency at which the rotor bars pass the flux detection point, and then it is synchronously time averaged. Averaging requires that corresponding portions of the flux signal, for a series of rotor revolutions, be added, and the synchronization signal is used as a reference for matching corresponding portions of the flux signal.

Abstract: A method and apparatus for detecting rotor faults in an induction motor. A flux sensor generates a flux signal corresponding to the magnetic flux at a predefined flux detection point external to the motor. A current sensor generates a current signal proportional to the current drawn by said motor. A time series of data points is stored, representing the values of the flux signal and the current signal over a period of time. The time series are transformed by FFT into a set of flux spectra and into a set of current spectra. Then the line frequency of the motor's power supply is determined by finding the maximum of the current spectra. Similarly, the slip frequency of the motor is determined by finding the maximum of the flux spectra in a predefined spectral range (e.g., below 2 Hz). The analysis of the rotor is then performed by comparing the amplitude of the current spectra, at a set of rotor fault harmonic frequencies, with specified fault threshold criteria.

Abstract: The quality of welds produced during a percussion welding process are determined in real time by sensing the vibrational impulse signals produced during the percussion welding process, filtering the impulse signal to select a frequency band containing frequency components associated with the vibrations imparted to the forge rod of the welding machine, measuring the mechanical damping characteristics of the selected spectral portion of the signal by determining the log-decrement of the envelope of the signal, and comparing the measured log-decrement value obtained with a preselected threshold value.

Abstract: Undesirable internal changes in a running turbine are diagnosed by monitoring the dynamic pressure of the stream of fluid and the structure-borne vibration of the stationary shell. Incipient erosion of blades, material deposit buildup, and other abnormalities are detected by analyzing the dynamic pressure and vibration signal utilizing several signal processing techniques. A better maintenance schedule can be derived, this also serves as a turbine efficiency monitor.