Implementation of a Fault-Diagnosis Algorithm for Induction Machines Based on Advanced Digital-Signal-Processing Techniques

In this paper, a complete cross-correlation-based fault-diagnostic method is proposed for real-time digital-signal-processor (DSP) applications that cover both the fault-monitoring and decision-making stages. In practice, a motor driven by an inverter or utility line is run at various operating points where the frequency, amplitude, and phase of the fault signatures vary unexpectedly. These changes are considered to be one of the common factors that yield erroneous fault tracking and unstable fault detection. In this paper, the proposed algorithms deal with the ambiguities of line-current noise or sensor-resolution errors and operating-point-dependent threshold issues. It is theoretically and experimentally verified that a motor fault can be continuously tracked when the sensor errors are within a limited range through the adaptively determined threshold definition of noise conditions. The offline experiments are performed via Matlab using actual line-current data obtained by a data-acquisition system. These results are verified on a DSP-based motor drive in real time where drive sensors and a digital signal processor are employed both for motor-control and fault-diagnostic purposes.