Detecting Faults in Turbine Generators

• US 20160033580A1
• Filed: 10/15/2015
• Published: 02/04/2016
• Est. Priority Date: 05/29/2012
• Status: Active Grant

• Current Assignee: NUtech Ventures
• Original Assignee: Board of Regents of the University of Nebraska
• Inventors: Qiao, Wei, Gong, Xiang
• Granted Patent: US 10,359,473 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: F03D 17/00 Monitoring or testing of wind motors, e.g. diagnostics testing during commissioning of wind motors F03D13/30
• CPC Class Codes: F03D 7/0296 to prevent, counteract or reduce noise emissions
• CPC Class Codes: F05B 2270/334 Vibration measurements
• CPC Class Codes: G01R 31/343 in operation
• CPC Class Codes: G01R 31/42 AC power supplies
• CPC Class Codes: H02P 29/02 Providing protection against overload without automatic interruption of supply protection against faults of stepper motors H02P8/36
• CPC Class Codes: Y02E 10/72 Wind turbines with rotation axis in wind direction

Abstract

Turbine generator faults can be detected by receiving a current signal from the turbine generator, and synchronously sampling the current signal to obtain a set of current signal samples that are evenly spaced in the phase domain, in which phase differences between adjacent current signal samples in the set of current signal samples are substantially the same. The process includes generating a frequency spectrum of the current signal samples, identifying one or more excitations in the frequency spectrum, and detecting a fault in the turbine generator based on the one or more excitations in the frequency spectrum.

Claims

1. A method for detecting turbine generator faults, the method comprising:

receiving a current signal from a turbine generator;

synchronously sampling the current signal to obtain a first set of current signal samples that are evenly spaced in the phase domain, in which phase differences between adjacent current signal samples in the first set are substantially the same;

generating a frequency spectrum of the first set of current signal samples;

identifying one or more excitations in the frequency spectrum; and

detecting a fault in the turbine generator based on the one or more excitations in the frequency spectrum.

2. The method of claim 1 in which detecting a fault in the turbine generator comprises detecting an excitation at a characteristic frequency that corresponds to a known type of turbine generator fault.

3. The method of claim 1, further comprising locally normalizing the frequency spectrum of the current signal samples, using a median filter to calculate median locally normalized values of the frequency spectrum, and

based on the median values, determining a threshold for detecting excitations in the frequency spectrum.

4. The method of claim 1 in which synchronously sampling the current signal comprises:

sampling the current signal according to a first set of sampling times to generate a second set of current signal samples,

determining instantaneous phases of the second set of current signal samples or current signal samples derived from the second set of current signal samples,

based on the instantaneous phases, calculating a second set of sampling times that correspond to phases of the current signal in which the phase intervals of adjacent samples are substantially constant, and

re-sampling the current signal according to the second set of sampling times to generate the first set of sampled current data.

5. The method of claim 4 in which receiving a current signal from a turbine generator comprises receiving an analog or digital current signal from the turbine generator, and re-sampling the current signal comprises re-sampling the analog or digital current signal according to the second set of sampling times.

6. The method of claim 4 in which re-sampling the current signal comprises calculating interpolated current signal amplitudes according to the second set of sampling times.

7. The method of claim 1 in which synchronously sampling the current signal comprises:

sampling the current signal according to a first set of sampling times to generate a second set of sampled current data,

applying a forward-backward filter and a Hilbert transform to determine amplitudes of normalized current signal,

determining instantaneous phases of the current signal based on the amplitudes of the normalized current signal, and

re-sampling the current signal based on information derived from the instantaneous phases.

8. The method of claim 1, comprising comparing the frequency spectrum of the first set of current signal samples with a reference frequency spectrum, in which detecting a fault in the turbine generator comprises detecting a difference between the frequency spectrum of the first set of current signal samples and the reference frequency spectrum.

9. The method of claim 1, further comprising, in response to detecting a fault in the turbine generator, generating and transmitting an alert that indicates a turbine generator fault is detected.

10. The method of claim 1 in which receiving a current signal from a turbine generator comprises receiving a current signal from at least one of a wind turbine generator, a hydro turbine generator, a steam turbine generator, or a gas turbine generator.

11. The method of claim 1 in which the current signal comprises a nonstationary periodic signal wherein the fundamental frequency of the current signal varies over time.

12. The method of claim 11 in which the current signal comprises a nonstationary sinusoidal signal.

13. The method of claim 1, further comprising normalizing amplitudes of the one or more excitations in the frequency spectrum,

calculating normalized power differences between normalized excitations determined at a current time and baseline normalized excitations, and

calculating standard deviations of the normalized power differences,

wherein detecting a fault in the turbine generator comprises detecting a fault in the turbine generator based on the standard deviations of the normalized power differences.

14. The method of claim 13, further comprising comparing the standard deviations of the normalized power differences with a threshold;

wherein detecting a fault in the turbine generator comprises detecting a fault in the turbine generator when the standard deviations of the normalized power differences is greater than the threshold.

15. The method of claim 13 in which detecting a fault in the turbine generator comprises detecting a gear fault in the turbine generator.

16. A controller configured to detect turbine generator faults, the controller having a processor that is configured to execute instructions to:

receive a current signal from a turbine generator;

synchronously sample the current signal to obtain a set of current samples that are evenly spaced in phase, in which phase differences between adjacent current samples in the set are substantially the same;

generate a frequency spectrum of the current signal samples;

identify one or more excitations in the frequency spectrum; and

detect a fault in the turbine generator based on the one or more excitations in the frequency spectrum.

17. The controller of claim 16 in which detecting a fault in the turbine generator comprises detecting a fault in the turbine generator by detecting an excitation at a characteristic frequency that corresponds to a known type of turbine generator fault.

18. The controller of claim 16 in which the processor is further configured to execute instructions to locally normalize the frequency spectrum of the current signal samples, use a median filter to calculate median locally normalized values of the frequency spectrum, and

based on the median values, determine a threshold for detecting excitations in the frequency spectrum.

19. The controller of claim 16 in which the processor is further configured to execute instructions to, in response to detecting a fault in the turbine generator, generate and transmit an alert that indicates that a turbine generator fault is detected.

20. The controller of claim 16 in which receiving a current signal from a turbine generator comprises receiving a current signal from at least one of a wind turbine generator, a hydro turbine generator, a steam turbine generator, or a gas turbine generator.

21. A system for detecting turbine generator faults, the system comprising:

at least one current sensor configured to sense a current signal from a turbine generator; and

a processor that is configured to execute instructions to;

receive the current signal from the at least one current sensor;

synchronously sample the current signal to obtain a set of current samples that are evenly spaced in phase, in which phase differences between adjacent current samples in the set are substantially the same;

generate a frequency spectrum of the current signal samples;

identify one or more excitations in the frequency spectrum; and

detect a fault in the turbine generator based on the one or more excitations in the frequency spectrum.

22. The system of claim 21 in which the turbine generator comprises at least one of a wind turbine generator, a hydro turbine generator, a steam turbine generator, or a gas turbine generator.