Method and apparatus for acoustic emissions testing

• US 9,989,501 B2
• Filed: 05/10/2016
• Issued: 06/05/2018
• Est. Priority Date: 05/10/2016
• Status: Active Grant

• Current Assignee: The Boeing Co.
• Original Assignee: The Boeing Co.
• Inventors: Tat, Hong Hue, Wu, Yuan-Jye, Schaefer, Joseph D., Mathews, Mary J., Kao, Anne, Pauca, Victor P., Li, Rongzhong
• Primary Examiner(s): Saint Surin, J M
• Application Number:
  US15/150,595
• Publication Number:
  US 20170328868A1
• Time in Patent Office: 756 Days
• Field of Search: 73587
• CPC Class Codes: G01H 1/00 Measuring characteristics of vibrations in solids by using direct conduction to the detector G01H9/00, G01H11/00 take precedence
• CPC Class Codes: G01L 1/106 Constructional details
• CPC Class Codes: G01N 29/14 using acoustic emission techniques echo of particles G01N29/046; measuring mechanical vibrations or acoustic waves in solids in general G01H1/00

Abstract

A method and apparatus for analyzing an object using acoustic emissions. Load data is received for the object. Acoustic waveform data is received for the object from an acoustic sensing system. The acoustic waveform data represents acoustic emissions emanating from the object and is detected using the acoustic sensing system. A plurality of bins is created for the load data. A plurality of frequency distribution functions is generated for the plurality of bins using the acoustic waveform data. A set of learning algorithms is applied to the plurality of frequency distribution functions and the acoustic waveform data to generate an output that allows an operator to more easily and quickly assess a structural integrity of the object.

Claims

1. An apparatus comprising:

an acoustic sensing system positioned relative to an object, wherein the acoustic sensing system detects acoustic emissions and generates acoustic waveform data for the acoustic emissions detected; and

an analyzer module implemented in a computer system that receives load data and the acoustic waveform data for the object, creates a plurality of bins for the load data, generates a plurality of frequency distribution functions for the plurality of bins using the acoustic waveform data, and applies a set of learning algorithms to the plurality of frequency distribution functions and the acoustic waveform data to generate an output that allows an operator to more easily and quickly assess a structural integrity of the object.

2. The apparatus of claim 1, wherein the acoustic sensing system comprises:

an acoustic sensor positioned in contact with the object.

3. The apparatus of claim 1, wherein the set of learning algorithms includes either a set of unsupervised learning algorithms or a set of supervised learning algorithms.

4. The apparatus of claim 1, wherein the load data is either retrieved from a database or received from a load sensing system that measures a loading of the object as the acoustic waveform data is generated.

5. The apparatus of claim 1, wherein the plurality of bins have a plurality of bin widths and wherein a bin width in the plurality of bin widths is either a time interval or a load interval.

6. The apparatus of claim 1, wherein each frequency distribution function in the plurality of frequency distribution functions has a plurality of bin widths in which each bin width of the plurality of bin widths is a defined frequency interval.

7. The apparatus of claim 1, wherein each frequency distribution function in the plurality of frequency distribution functions comprises a plurality of frequency bins and wherein a frequency bin in the plurality of frequency bins includes either a count of a number of frequency peaks that fall within the frequency bin or an accumulation of energy at the frequency bin computed using the acoustic waveform data.

8. The apparatus of claim 1, wherein the output identifies a plurality of clusters of the plurality of frequency distribution functions and wherein the plurality of clusters comprises:

a first cluster representing a first mode of structural change;

a second cluster representing a second mode of structural change;

a third cluster representing a third mode of structural change; and

a fourth cluster representing a fourth mode of structural change.

9. The apparatus of claim 1, wherein the analyzer module applies the set of learning algorithms to the plurality of frequency distribution functions to establish a plurality of clusters and to identify a plurality of descriptors for the plurality of clusters.

10. The apparatus of claim 9, wherein the plurality of clusters is analyzed with alternate test data to associate each descriptor in the plurality of descriptors with a different mode of structural change and wherein the alternate test data is selected from one of x-ray imaging data, ultrasound imaging data, infrared imaging data, and modeling data.

11. The apparatus of claim 10, wherein the analyzer module generates a descriptor classification output that associates a mode of structural change with the each descriptor in the plurality of descriptors and wherein the descriptor classification output is stored in a database for future use in evaluating a structural integrity of a part during at least one stage in a lifecycle of the part.

12. A method for analyzing an object using acoustic waves, the method comprising:

receiving load data for the object;

receiving acoustic waveform data for the object from an acoustic sensing system, wherein the acoustic waveform data represents acoustic emissions emanating from the object and is detected using the acoustic sensing system;

creating a plurality of bins for the load data;

generating a plurality of frequency distribution functions for the plurality of bins using the acoustic waveform data; and

applying a set of learning algorithms to the plurality of frequency distribution functions and the acoustic waveform data to generate an output that allows an operator to more easily and quickly assess a structural integrity of the object.

13. The method of claim 12 further comprising:

detecting, by the acoustic sensing system, acoustic waves radiating from the object using at least one acoustic sensor to generate an acoustic emissions signal; and

converting the acoustic emissions signal into the acoustic waveform data.

14. The method of claim 12, wherein, creating the plurality of bins comprises:

identifying a plurality of bin widths for the plurality of bins, wherein a bin width in the plurality of bin widths is either a defined time interval or a defined load interval; and

identifying a set of waveforms in the acoustic waveform data that fall within each bin of the plurality of bins.

15. The method of claim 14, wherein generating the plurality of frequency distribution functions comprises:

dividing a selected frequency range into a plurality of frequency bins based on at least one defined frequency interval.

16. The method of claim 15, wherein generating the plurality of frequency distribution functions further comprises:

computing a Fast Fourier Transform for the set of waveforms that fall within a bin selected from the plurality of bins; and

updating the plurality of frequency bins in a frequency distribution function for the bin based on the Fast Fourier Transform.

17. The method of claim 16, wherein updating the plurality of frequency bins comprises:

selecting a number of frequency peaks for each waveform in the set of waveforms; and

incrementing a frequency bin in the plurality of frequency bins when a frequency peak in the number of frequency peaks has a frequency that falls within the frequency bin.

18. The method of claim 12, wherein applying the set of learning algorithms comprises:

applying a set of unsupervised learning algorithms to the plurality of frequency distribution functions and the acoustic waveform data to establish a plurality of clusters;

identifying a plurality of descriptors for the plurality of clusters; and

generating a descriptor classification output that classifies each descriptor in the plurality of descriptors as representing a different mode of structural change based on a plurality of modes identified using alternate test data.

19. The method of claim 12, wherein applying the set of learning algorithms comprises:

applying a set of supervised learning algorithms to the plurality of frequency distribution functions, the acoustic waveform data, and a stored plurality of descriptors; and

generating a classification result for each frequency distribution function of the plurality of frequency distribution functions.

20. A method for monitoring a composite object in an aircraft during at least one stage in a lifecycle of the aircraft, the method comprising:

detecting acoustic emissions radiating from the composite object using an acoustic sensing system to generate acoustic waveform data;

receiving, by an analyzer module, the acoustic waveform data and load data for the composite object;

creating, by the analyzer module, a plurality of bins for the load data, wherein a set of waveforms in the acoustic waveform data falls within a corresponding bin in the plurality of bins;

generating, by the analyzer module, a plurality of frequency distribution functions for the plurality of bins using the acoustic waveform data; and

applying a set of supervised learning algorithms to the plurality of frequency distribution functions, the acoustic waveform data, and a stored plurality of descriptors to generate a classification output that identifies a classification result for each frequency distribution function in the plurality of frequency distribution functions and in which the classification output allows an operator to more easily and quickly assess a structural integrity of the composite object.

21. The method of claim 20, wherein applying the set of supervised learning algorithms to the plurality of frequency distribution functions, the acoustic waveform data, and the stored plurality of descriptors to generate the classification output comprises:

generating the classification result for a selected frequency distribution function in the plurality of frequency distribution functions by comparing the selected frequency distribution function to each of the stored plurality of descriptors, wherein the classification result identifies whether the selected frequency distribution function, and thereby a corresponding set of waveforms in the acoustic waveform data, represents zero or more modes of structural change.