A park transform-based method for condition monitoring of three-phase electromechanical systems

This paper presents a Park transform-based method for preprocessing stator current data from a motor and transforming it into a form that is useful for fault detection and diagnostics. The proposed method generates power signatures that are invariant to the initial electrical angle of the voltage when the motor is connected to the utility, and can also adapt to variations in the electrical angle of the supply voltage over time. A modified nonlinear least squares algorithm identifies and tracks the parameters of the supply voltage over time, ensuring that the supply voltage and the argument of the Park transformation remain synchronized. Experimental results are presented that illustrate the method's effectiveness for identifying changes in the mechanical load on a 3/4 HP refrigeration compressor.