System and method for motor fault detection using stator current noise cancellation

1. A controller configured to detect indicia of incipient mechanical motor faults having a processor programmed to:

receive a first set of real-time operating current data from a motor during operation;

define a noise component present in the first set of real-time operating current data;

remove the noise component from the first set of real-time operating current data to identify any fault components present in the first set of real-time operating current data;

generate a fault index for the first set of real-time operating current data based on any isolated fault components;

acquire at least one additional set of real-time operating current data from the motor during operation;

redefine the noise component present in each of the at least one additional sets of real-time operating current data;

remove the redefined noise component from each of the at least one additional sets of real-time operating current data to identify any fault components present therein; and

generate a fault index for each of the at least one additional sets of real-time operating current data based on any isolated fault components.

2. The controller of claim 1 wherein the processor is further programmed to redefine the noise component present in each of the at least one additional sets of real-time operating current data based on periodic components therein.

3. The controller of claim 2 wherein the periodic components in each of the at least one additional sets of real-time operating current data comprise sinusoidal signals of varying frequency and amplitude.

4. The controller of claim 1 wherein the processor is further programmed to apply a statistical process control analysis to a plurality of generated fault indices to calculate a fault index warning threshold.

5. The controller of claim 4 wherein the processor is further programmed to generate an alert if a detected variation of the fault index exceeds the fault index warning threshold.

6. The controller of claim 5 wherein the processor is further programmed to:

determine a percentage of fault indices that exceed the fault index warning threshold; and

generate the alert if the percentage of fault indices that exceed the fault index warning threshold exceed a pre-determined percentage.

7. The controller of claim 1 wherein the processor is further programmed to:

configure a Wiener filter based on the first set of real-time operating current data; and

reconfigure the Wiener filter based on each of the at least one additional sets of real-time operating current data, the Wiener filter configured to estimate the redefined noise component corresponding to the first set of real-time operating current data and each additional set of real-time operating current data.

8. The controller of claim 7 wherein the fault components comprise a prediction error of the Wiener filter.

9. The controller of claim 1 wherein the processor is further programmed to calculate a root-mean-square (RMS) value of the fault components to generate the fault index.

10. The controller of claim 1 wherein the fault components comprise non-periodic components in each set of real-time operating current data.

11. A non-invasive method for detecting impending faults in electric machines comprising:

acquiring a plurality of stator current data sets from the electric machine during operation;

configuring a current data filter for each of the plurality of stator current data sets;

applying each of the current data filters to its respective stator current data set in real-time to generate a noise-cancelled stator current;

determining a fault index from the noise-cancelled stator current for each of the plurality of stator current data sets;

monitoring a value of the fault index for each of the plurality of stator current data sets; and

generating an alert if the value of a pre-determined number of fault indices exceeds a control limit.

12. The method of claim 11 wherein configuring the current data filter for each of the plurality of stator current data sets comprises reconfiguring a Wiener filter for each of the plurality of stator current data sets.

13. The method of claim 12 wherein reconfiguring the Wiener filter for each of the plurality of stator current data sets comprises estimating a periodic noise component from each of the stator current data sets.

14. The method of claim 13 wherein generating the noise-cancelled stator current for each of the plurality of stator current data sets comprises cancelling the periodic noise components from each of the plurality of stator current data sets.

15. The method of claim 12 further comprising selecting coefficients in each reconfigured Wiener filter by applying a minimum mean-squared error (MMSE) operation to the stator current data set associated with the reconfigured Wiener filter.

16. A system for monitoring current to predict faults comprising:

at least one non-invasive current sensor configured to acquire stator current data from an operating motor; and

a processor connected to receive the stator current data from the at least one non-invasive current sensor, the processor programmed to:

repeatedly receive a set of real-time operating current data from the at least one non-invasive current sensor, the set of real-time operating data representative of real-time motor operation;

define a non-fault component from each of the repeatedly received sets of real-time operating current data, the non-fault component being a periodic component of the real-time operating current data;

remove the non-fault component from each of the sets of real-time operating current data in real-time to isolate residual current data;

process the residual current data to identify possible bearing faults;

generate a fault index for any identified bearing faults; and

generate an alert if the fault index exceeds a fault index threshold.

17. The current monitoring system of claim 16 wherein the processor is further programmed to repeatedly reconfigure a Wiener filter based on each of the sets of real-time operating current data, the Wiener filter reconfigured to redefine the non-fault component in each of the sets of real-time operating current data.

18. The current monitoring system of claim 17 wherein the processor is further programmed to reconfigure the Wiener filter to define the periodic component in each of the sets of real-time operating current data.