Apparatus and method for detecting the current damaged state of a machine

Patent metadata

Field	Value
Publication number	US-10088830-B2
Application number	US-201415102568-A
Country	US
Kind code	B2
Filing date	Oct 28, 2014
Priority date	Dec 16, 2013
Publication date	Oct 2, 2018
Grant date	Oct 2, 2018

Abstract

A method and apparatus for detecting the current damaged state of a machine is provided. The disclosed relates to an apparatus for detecting the current damaged state of a machine, wherein the data capture and editing device: comprises a Fast Fourier analysis apparatus, for performing a transformation from pressure signals that are available in the time domain to a frequency domain as discrete-time frequency spectra ( 2 ), additionally includes a spatial aggregation unit and a temporal aggregation unit that combines the discrete-time frequency spectra, additionally includes a cepstrum analysis unit, for producing a cepstrum with various actual cepstrum values from the temporally and spatially aggregated discrete frequency spectra ( 5 ), additionally includes a computation unit that takes the cepstra produced in this manner and produces, at a previously defined interval of time, a plurality of actual cepstrum intervals including adjacent actual cepstrum values, as a result of which an actual cepstrum value, which can be ascertained from at least three adjacent actual cepstrum intervals, can be produced ( 6 ), includes a unit that computes ( 7 ) a target/actual comparison with the actual cepstrum value produced in this manner and a previously determined target cepstrum value.

First claim

The invention claimed is:

1. A device for detecting the current state of damage of a machine, comprising:
a plurality of pressure sensors for detecting the pressure over a time period of the machine to be monitored and outputting a plurality of pressure signals, wherein the pressure signals exist in the time domain, and a data detection and processing apparatus for detecting the pressure signals from the pressure sensors and processing the pressure signals,
wherein:
the data detection and processing apparatus comprises a fast Fourier analysis device for carrying out a transformation of the pressure signals existing in the time domain into a frequency domain as discrete-time frequency spectra;
the data detection and processing apparatus furthermore comprises a spatial aggregation unit and a temporal aggregation unit, which combine the discrete-time frequency spectra, such that temporally and spatially aggregated discrete frequency spectra are generable, wherein the application of the spatial aggregation unit and also the application of the temporal aggregation unit is performable in a different sequence on the discrete-time frequency spectra;
the data detection and processing apparatus moreover includes a cepstrum analysis unit for generating a cepstrum with different actual cepstra values from the temporally and spatially aggregated discrete frequency spectra;
wherein the data detection and processing apparatus moreover includes a computer unit, which, from the cepstra generated, generates actual cepstra intervals in a predefined time interval, said actual cepstra intervals comprising a plurality of adjacent actual cepstra values, as a result of which an actual cepstrum value, which is establishable from at least three adjacent actual cepstra intervals, is generable; and
wherein the data detection and processing apparatus comprises a unit which calculates a comparison of the actual cepstrum value generated and an intended cepstrum value determined in advance.

2. The device for detecting the current state of damage of a machine as claimed in claim 1 ,
wherein:
the spatial aggregation unit combines the discrete-time frequency spectra to form spatially virtual frequency spectra and subsequently the temporal aggregation unit combines the spatially virtual frequency spectra to temporally-spatially virtual target frequency spectra; and
and subsequently the cepstrum analysis unit generates a cepstrum with different actual cepstra values from the temporally-spatially virtual target frequency spectrum.

3. The device for detecting the current state of damage as claimed in claim 1 wherein the data detection and processing apparatus generates the intended cepstrum with the aid of an undamaged machine, from which the intended cepstrum value agreeing with the actual cepstrum value is then determinable.

4. The device for detecting the current state of damage as claimed in claim 1 , wherein the actual cepstrum value is correlatable with the degree of damage of the monitoring machine.

5. The device for detecting the current state of damage as claimed in claim 1 , wherein a machine unit is provided, which initiates an emergency operation if a predefined threshold of the comparison is exceeded.

6. The device for detecting the current state of damage as claimed in claim 1 , wherein the plurality of pressure sensors are dynamic pressure sensors.

7. The device for detecting the current state of damage as claimed in claim 1 , wherein the machine is a compressor with rotor blades and a rotor.

8. The device for detecting the current state of damage as claimed in claim 7 ,
wherein the compressor comprises a compressor circumference and a compressor outlet and the plurality of pressure sensors are attached over the compressor circumference at the compressor outlet.

9. The device for detecting the current state of damage as claimed in claim 7 , wherein the compressor comprises a compressor circumference and the pressure sensors are not attached below these rotor blades on the compressor circumference.

10. The device for detecting the current state of damage as claimed in claim 7 , wherein the rotor has a rotary frequency and the minimum bandwidth of the data detection and processing apparatus emerges from the maximum number of rotor blades multiplied by twice the rotary frequency of the rotor.

11. The device for detecting the current state of damage as claimed in claim 1 , wherein the machine has a compressor outlet collection chamber and a burner region, and the pressure sensors are arranged in at least one of the compressor outlet collection chamber and the burner region.

12. The device for detecting the current state of damage as claimed in claim 1 , wherein the actual cepstrum value, which is establishable from at least three adjacent actual cepstra intervals, is generable by summation.

13. A method for detecting the current state of damage of a machine, in which a pressure over a time period of the machine to be monitored is detected by way of a plurality of pressure sensors, and a plurality of pressure signals, which exist in the time domain, are output comprising the following steps:
transforming the pressure signals existing in the time domain into the frequency domain as discrete-time frequency spectra by a fast Fourier analysis device provided in a data detection and processing apparatus;
spatially and temporally combining the discrete-time frequency spectra into temporally and spatially aggregated discrete frequency spectra by a spatial aggregation unit and temporal aggregation unit provided in the data detection and processing apparatus;
generating a cepstrum with different actual cepstra values from the temporally and spatially aggregated discrete frequency spectra by a cepstrum analysis unit;
generating, from the cepstra generated, actual cepstra intervals comprising at least a plurality of adjacent actual cepstra values in a predefined time interval by a computer unit;
generating an actual cepstrum value which is established from at least three adjacent actual cepstra intervals; and
calculating a comparison of the actual cepstrum value generated and an intended cepstrum value determined in advance.

14. The method for detecting the current state of damage of a machine as claimed in claim 13 ,
wherein:
the discrete-time frequency spectra are combined to spatially virtual frequency spectra in the spatial aggregation unit;
subsequently, the spatially virtual frequency spectra are combined to temporally-spatially virtual target frequency spectra in the temporal aggregation unit;

subsequently, a cepstrum with different actual cepstra values is generated from the temporally-spatially virtual target frequency spectrum in the cepstrum analysis unit.

15. The method as claimed in claim 13 , wherein
temporal combining of the spatially virtual frequency spectra or the discrete-time frequency spectra is carried out using the peak-hold algorithm or by forming an average.

16. The method for detecting as claimed in,
claim 13 , wherein the actual cepstrum value is generated from at most nine adjacent actual cepstra intervals.

17. The method for detecting the current state of damage as claimed in 6 , claim 13 , wherein the actual cepstrum value, which is establishable from at least three adjacent actual cepstra intervals, is generated by summation.

18. The method for detecting as claimed in,
claim 13 , wherein the actual cepstrum value correlates with the degree of damage of the machine to be monitored.

19. The method for detecting as

Assignees

• Siemens Ag

Inventors

• Brummel Hans-Gerd
• Pfeifer Uwe

Classifications

• G05B2219/37399

  Pressure · CPC title

• G05B23/024 Primary

  Quantitative history assessment, e.g. mathematical relationships between available data; Functions therefor; Principal component analysis [PCA]; Partial least square [PLS]; Statistical classifiers, e.g. Bayesian networks, linear regression or correlation analysis; Neural networks · CPC title

• G05B2219/37371

  Flow · CPC title

• G05B19/416

  characterised by control of velocity, acceleration or deceleration (G05B19/19 takes precedence) · CPC title

• G01H1/006 Primary

  of the rotor of turbo machines · CPC title