Method and system for the vibration analysis of an engine

• US 10,281,360 B2
• Filed: 05/20/2014
• Issued: 05/07/2019
• Est. Priority Date: 05/17/2013
• Status: Active Grant

• Current Assignee: Safran Aircraft Engine Company
• Original Assignee: Safran Aircraft Engine Company
• Inventors: Lacaille, Jerome
• Primary Examiner(s): Caputo, Lisa M
• Assistant Examiner(s): Hernandez-Prewitt, Roger G
• Application Number:
  US14/890,472
• Publication Number:
  US 20160103038A1
• Time in Patent Office: 1,813 Days
• Field of Search: 7311407, 707705, 707713
• CPC Class Codes: G01M 15/12 by monitoring vibrations
• CPC Class Codes: G05B 23/0229 knowledge based, e.g. expert systems; genetic algorithms

Abstract

A method and a system for the vibration analysis of an engine, including acquiring vibration signals relating to the engine and in order to form at least one spectrogram that represents an operating state of the engine, selecting at least one zone of interest in at least one spectrogram, comparing each zone of interest with a set of corresponding comparison zones belonging to annotated spectrograms recorded in a reference database, and determining a subset of comparison zones that have vibration behaviors similar to those of the zone of interest, and displaying the subset of comparison zones and annotations associated with the subset of comparison zones.

Claims

1. A method for the vibration analysis of an engine, comprising:

measuring, by vibration sensors, vibrations of the engine during acceleration and deceleration phases according to low-pressure and high-pressure rotation speeds;

spectrally acquiring, by a processor, vibration signals from the measuring of the vibrations, the vibration signals relating to the engine in order to form at least one spectrogram that represents an operating state of said engine;

selecting at least one zone of interest in said at least one spectrogram;

constructing for each zone of interest a vector of interest that represents said zone of interest by projecting said zone of interest onto an underlying representation space;

constructing a set of comparison vectors that is representative of a set of corresponding comparison zones with said zone of interest by projecting each one of said comparison zones onto said representation space, said set of comparison zones belonging to annotated spectrograms recorded in a reference database;

calculating a similarity score between said vector of interest and each one of said comparison vectors for determining a subset of comparison vectors that have a similarity score that exceeds a predetermined level;

selecting a subset of comparison zones associated to said subset of comparison vectors, said subset of comparison zones having vibration behaviours similar to those of said zone of interest; and

recovering said subset of comparison zones and annotations associated with said subset of comparison zones.

2. The method according to claim 1, further comprising:

annotating said zone of interest; and

recording in the reference database of said zone of interest, the annotation associated to said zone of interest as well as the spectrogram containing said zone of interest.

3. The method according to claim 1, wherein projecting onto the representation space is carried out according to a projection model or a combination of projection models from among the following models:

physical model, learning model, and mathematical model.

4. The method according to claim 1, further comprising displaying a list of said subset of comparison zones diagnosed as being closest to the zone of interest.

5. The method according to claim 4, wherein said list of said subset of comparison zones is sorted in decreasing order of similarity and comprises statistical data and annotations.

6. The method according to claim 1, further comprising cleaning said at least one spectrogram or said at least one zone of interest with a noise removal algorithm.

7. The method according to claim 6, wherein the noise removal algorithm corresponds to a parametric model of a Gamma law, a statistical model, or a phase shift model.

8. The method according to claim 1, wherein each comparison zone is included in a spectral domain that corresponds to a vicinity of the zone of interest.

9. The method according to claim 1, wherein the engine is an aircraft engine.

10. A system for vibration analysis of an engine, said system comprising:

a processor configured to;

acquire, from vibration sensors that measure vibrations of the engine during acceleration and deceleration phases according to low-pressure and high-pressure rotation speeds, vibration signals relating to the engine in order to form at least one spectrogram that represents an operating state of said engine,

select at least one zone of interest in said at least one spectrogram,

construct for each zone of interest a vector of interest that represents said zone of interest by projecting said zone of interest onto an underlying representation space,

construct a set of comparison vectors that represent a set of corresponding comparison zones with said zone of interest by projecting each one of said comparison zones onto said representation space, said set of comparison zones belonging to annotated spectrograms recorded in a reference database,

calculate a similarity score between said vector of interest and each one of said comparison vectors in order to determine a subset of comparison vectors that have a similarity score that exceeds a predetermined level,

select a subset of comparison zones associated with said subset of comparison vectors, said subset of comparison zones having vibration behaviours similar to those of said zone of interest, and

display said subset of comparison zones and annotations associated with said subset of comparison zones.