Methods for monitoring structural health conditions

• US 20050075846A1
• Filed: 09/16/2004
• Published: 04/07/2005
• Est. Priority Date: 09/22/2003
• Status: Active Grant

• Current Assignee: Advanced Structure Monitoring, Inc.
• Original Assignee: Advanced Structure Monitoring, Inc.
• Inventors: Kim, Hyeung-Yun
• Granted Patent: US 7,286,964 B2
• Time in Patent Office: Days
• US Class Current: 703/1
• CPC Class Codes: G01H 11/00 Measuring mechanical vibrations or ultrasonic, sonic or infrasonic waves by detecting changes in electric or magnetic properties
• CPC Class Codes: G01H 11/08 using piezoelectric devices
• CPC Class Codes: G01H 9/004 using fibre optic sensors light guides per se G02B6/00, acousto-optical devices specially adapted for gating or modulating in optical wave guides G02F1/125
• CPC Class Codes: G01M 11/086 Details about the embedment of the optical fiber within the DUT
• CPC Class Codes: G01M 5/0033 by determining damage, crack or wear
• CPC Class Codes: G01M 5/0066 by exciting or detecting vibration or acceleration vibration testing of structures G01M7/00
• CPC Class Codes: G01M 5/0091 by using electromagnetic excitation or detection
• CPC Class Codes: G01N 2291/011 Velocity or travel time
• CPC Class Codes: G01N 2291/015 Attenuation, scattering
• CPC Class Codes: G01N 2291/02491 Materials with nonlinear acoustic properties
• CPC Class Codes: G01N 2291/0258 Structural degradation, e.g. fatigue of composites, ageing of oils
• CPC Class Codes: G01N 2291/0422 Shear waves, transverse waves, horizontally polarised waves
• CPC Class Codes: G01N 2291/0423 Surface waves, e.g. Rayleigh waves, Love waves
• CPC Class Codes: G01N 2291/0427 Flexural waves, plate waves, e.g. Lamb waves, tuning fork, cantilever
• CPC Class Codes: G01N 2291/106 one or more transducer arrays
• CPC Class Codes: G01N 29/043 in the interior, e.g. by shear waves
• CPC Class Codes: G01N 29/0609 Display arrangements, e.g. colour displays indicating or recording in connection with measuring in general G01D
• CPC Class Codes: G01N 29/0672 by acoustic tomography medical tomography A61B8/13
• CPC Class Codes: G01N 29/07 by measuring propagation velocity or propagation time of acoustic waves
• CPC Class Codes: G01N 29/223 Supports, positioning or alignment in fixed situation mounting transducers per se G10K11/004
• CPC Class Codes: G01N 29/245 : Ceramic probes, e.g. lead zirconate titanate [PZT] probes
• CPC Class Codes: G01N 29/2475 : Embedded probes, i.e. probes incorporated in objects to be inspected
• CPC Class Codes: G01N 29/2493 : Wheel shaped probes
• CPC Class Codes: G01N 29/28 : providing acoustic coupling , e.g. water impedance matching G10K11/02
• CPC Class Codes: G01N 29/4418 : with a model, e.g. best-fit, regression analysis
• CPC Class Codes: G01N 29/4463 : Signal correction, e.g. distance amplitude correction [DAC], distance gain size [DGS], noise filtering

Abstract

Methods and recordable media for monitoring structural health conditions. The present invention provides a method for interrogating a damage of a host structure using a diagnostic network patch (DNP) system having patches. An interrogation module partitions the plurality of patched in subgroups and measures the sensor signals generated and received by actuator and sensor patches, respectively. Then, a process module loads sensor signal data to identify Lamb wave modes, determine the time of arrival of the modes and generate a tomographic image. It also determines distribution of other structural condition indices to generate tomographic images of the host structure. A set of tomographic images can be stacked to generate a hyperspectral tomography cube. A classification module generates codebook based on K-mean/Learning Vector Quantization algorithm and uses a neural-fuzzy-inference system to determine the type of damages of the host structure.

Claims

1. A method for interrogating damages, identifying impacts and monitoring curing and repaired-boning-patch performance of a composite structure using a diagnostic network patch (DNP) system that is implemented thereto and comprises a plurality of patches, said method comprising:

partitioning the plurality of patches into one or more subgroups, each of the one or more subgroups having at least one actuator patch and at least one sensor patch;

designing a network and a plurality of signal paths using a genetic algorithm;

generating a signal by activating a first one of the plurality of patches;

receiving the generated signal via a second one of the plurality of patches through a corresponding one of the plurality of signal paths;

comparing the received signal with a baseline signal to interrogate the damage, the baseline signal measured in absence of the damages; and

storing the received signal and deviation of the received signal from the baseline signal.

2. The method of claim 1, further comprising, prior to the step of storing;

converting the baseline signal and received signal into extensible Markup Language (XML) formatted data.

3. The method of claim 1, wherein the signal is a Lamb wave signal.

4. The method of claim 1, further comprising:

storing coordinates of the plurality of patches and setup information.

5. The method of claim 4, wherein the setup information includes an actuation frequency, types and identification numbers of the patches, a voltage level and an operational failure status.

6. A method for identifying Lamb wave modes and determining time of arrivals of the Lamb wave modes, comprising:

loading a set of sensor signal data, each sensor signal data comprising Lamb wave signals measured at an excitation frequency;

detrending each of the set of sensor signal data to remove non-stationary signal component;

removing an electrical noise due to a toneburst actuator signal by applying a masking window to the each detrended sensor signal data;

performing a transformation on the each noise-removed sensor signal data to obtain a time-frequency signal energy distribution;

generating a multi-bandwidth energy distribution on a time-frequency plane by accumulating entire set of time-frequency signal energy distributions;

extracting one or more ridges from the multi-bandwidth energy distribution; and

identifying Lamb wave modes and determining time of arrivals of the Lamb wave modes based on the extracted one or more ridges.

7. The method of claim 6, wherein the Lamb wave modes include S₀, S_0—

_ref and A₀ modes.

8. The method of claim 6, wherein the transformation is a short-time Fourier transformation or a wavelet transformation.

9. A method for generating structural condition index (SCI) datasets from a plurality of sensor signal datasets, comprising:

(a) loading a plurality of sensor signal datasets for a plurality of network paths, each of the plurality of sensor signal datasets measured at one excitation frequency;

(b) selecting one of the plurality of sensor signal datasets;

(c) selecting one sensor signal from the selected sensor signal dataset, wherein the selected sensor signal is a Lamb wave signal;

(d) detrend and partition the selected sensor signal by applying an average filter and a masking window, respectively;

(e) decomposing the partitioned sensor signal into sub-bandwidth wave packets by applying a wavelet decomposition filter;

(f) synthesizing new sub-bandwidth packets;

(g) extracting S₀, S<sub>0—

ref</sub>, and A₀ modes from the synthesized sub-bandwidth packets by applying a set of envelop windows;

(h) computing at least one parameter of the set of envelop windows;

(i) determining a structural condition index for the selected sensor signal;

(j) determining a first discrete probability distribution function (DPDF) of the selected sensor signal;

(k) calculating a normality constant of the selected sensor signal;

(I) repeating the steps (d)-(k) for each sensor signal of the selected sensor signal dataset;

(m) determining a second DPDF for SCI dataset comprising structural condition indices obtained at step (i);

(n) finding and removing one or more outliers of the second DPDF;

(o) compensating an effect of ambient temperature on the SCI dataset;

(p) storing the SCI dataset; and

(q) repeating the steps (c)-(p) for each of the plurality of sensor signal datasets.

10. The method of claim 9, wherein the selected sensor signal comprises information of sensor coordinates.

11. The method of claim 9, further comprising, prior to the step of storing;

converting the SCI dataset into extensible Markup Language (XML) formatted data.

12. The method of claim 9, wherein the wavelet decomposition filter is a Daubechies wavelet filter.

13. The method of claim 9, wherein the SCI determined at step (i) is based on a change in spectrum energy of one of the S₀, S_0—

_ref and A₀ modes.

14. The method of claim 9, wherein the SCI determined at step (i) is based on a change in time of arrival of one of the S₀, S_0—

_ref and A₀ modes, maximum values of the envelop windows or center positions of the envelope windows.

15. The method of claim 9, wherein the SCI determined at step (i) is based on a structural dynamic parameter including natural frequencies, damping ratios and mode shapes of structures.

16. The method of claim 15, wherein the structural dynamic parameter change is obtained from vibration signals of an accelerometer, a displacement transducer or a strain gauge.

17. A method for generating a tomographic image to identify a region having changes in structural conditions that include damages of a host structure, comprising:

(a) loading a coordinate data of a plurality of diagnostic patches and a set of structural condition index (SCI) values for network paths defined by the plurality of diagnostic patches, the set of SCI values measured at an excitation frequency;

(b) calculating a bisection point for each of the network paths and assigning a corresponding one of the set of SCI values to the bisection point;

(c) calculating intersection points of the network paths;

(d) designating a SCI product to each of the intersection points;

(e) calculating SCI values near the intersection points using 3-dimensional SCI Gaussian functions, each of the 3-dimensional Gaussian functions defined for each of the network paths;

(f) generating a SCI distribution over a network plane by interpolation and a set of mesh-grid points of the network plane;

(g) setting up a chromosome population by assigning each chromosome to a corresponding one of the mesh-grid points;

(h) evaluating and ranking the chromosome population;

(i) selecting parent chromosomes from the chromosome population and reproducing child chromosomes;

(j) replacing the parent chromosomes with the reproduced child chromosomes;

(k) repeating the steps (i)-(j) over a preset number of generations to generate a final population of chromosomes;

(l) refining the SCI distribution over the final population of chromosomes; and

(m) generating a tomographic image of the refined SCI distribution.

18. The method of claim 17, further comprising, after the step of generating a tomographic image;

converting the refined SCI distribution into extensible Markup Language (XML) formatted data.

19. The method of claim 17, wherein the interpolation uses Delaunay triangulation technique.

20. The method of claim 17, further comprising:

repeating the steps (a)-(m) to generate a set of tomographic images, each of the set of tomographic image based on a set of SCI values measured at a corresponding excitation frequency; and

stacking the set of tomographic images to generate a hyperspectral tomographic cube.

21. A method for generating a tomographic image to identify changes in structural conditions that include damages of a host structure, comprising:

(a) loading a time of arrival dataset of a Lamb wave mode for a plurality of network paths defined by a plurality of diagnostic patches, the plurality of diagnostic patches applied to the host structure;

(b) applying an algebraic reconstruction technique to reconstruct the loaded time of arrival dataset; and

(c) generating a tomographic image of entire region of the host structure based on the reconstructed dataset.

22. The method of claim 21, wherein the mode is S₀ mode.

23. The method of claim 21, further comprising:

repeating the steps (a)-(c) to generate a set of tomographic images of a region, each of the set of tomographic images based on a time of arrival dataset measured at an excitation frequency; and

stacking the set of tomographic images to generate a hyperspectral tomographic cube of the region.

24. The method of claim 21, further comprising:

(d) rearranging the plurality of network paths to focus on a suspicious subregion;

(e) applying a simultaneous iterative construction technique to reconstruct iteratively the loaded time of arrival dataset; and

(f) generating a tomographic image of the suspicious subregion based on the iteratively reconstructed dataset.

25. The method of claim 23, further comprising:

repeating the steps (a)-(f) to generate a set of tomographic images of the suspicious region, each tomographic image of the suspicious region based on a time of arrival dataset measured at a corresponding excitation frequency; and

stacking the set of tomographic images of the suspicious region to generate a hyperspectral tomographic cube of the suspicious region.

26. A method for developing a codebook that is utilized to classify types of damages in a structure, comprising:

(a) initializing a set of cluster centers by randomly selecting a plurality of structural condition index (SCI) values on a plurality of grid points;

(b) determining a membership matrix;

(c) compute a cost function;

(d) updating the set of cluster centers;

(e) repeating the steps (b)-(d) if the cost is greater than a tolerance and the cost decreases upon repetition of the steps (b)-(d);

(f) labeling the set of cluster enters by a voting method;

(g) selecting a training SCI input vector randomly and choosing one of the cluster centers that is closest to the training SCI input vector;

(h) updating the chosen cluster center if the SCI input vector and the chosen cluster center belong to a same class; and

(i) generating a codebook including the updated cluster center.

27. The method of claim 26, further comprising:

building a damage classifier or an impact classifier using the generated codebook.

28. The method of claim 27, further comprising:

refining the generated codebook using a principal component analysis incorporated with Fisher linear discriminant analysis or eigenspace separation transformation.

29. The method of claim 28, wherein each of the plurality of structural condition index (SCI) values is a wavelet-transformed SCI value.

30. A method for developing a prognosis model to forecast damage evolution in a structure, comprising:

(a) building an input-output system model for a Lamb-wave network system having at least one actuator and at least one sensor at a selected time step;

(b) identifying the input-output system model using a state space system identification method;

(c) training a previous input-output system model with SCI values to generate an one-step-ahead system model using a recurrent neural network, the SCI values provided from an input-output system model built at a previous time step;

(d) generating output signals from the one-step-ahead system model using input signals measured by the at least one sensor;

(e) computing SCI values from the output signals;

(f) repeating the steps (c)-(e) until the iteration steps reaches a preset time of forecasting damage of the structure;

(g) generating future output signals of a future system model using the input signals, wherein the future system model is an input-output system model built at the preset time of forecasting damage; and

(h) providing SCI values of the future output signals.

31. A method for generating a 3-dimensional damage evolution manifold, comprising:

providing a host structure;

making a set of 2-dimensional tomographic images, each of the set of tomographic images generated after a preset number of vibrational repetitions are applied to the host structure;

stacking the tomographic images in an increasing order of vibrational repetitions to generate a 3-dmensional damage evolution manifold.

32. A computer readable medium carrying one or more sequences of instructions for interrogating a damage of a structure using a diagnostic network patch (DNP) system implemented thereto, the DNP system comprising a plurality of patches, wherein execution of one or more sequences of instructions by one or more processors causes the one or more processors to perform the steps of:

partitioning the plurality of patches into one or more subgroups, each of the one or more subgroups having at least one actuator patch and at least one sensor patch;

designing a network and a plurality of signal paths using a genetic algorithm;

generating a signal by activating a first one of the plurality of patches;

receiving the generated signal via a second one of the plurality of patches through a corresponding one of the plurality of signal paths;

comparing the received signal with a baseline signal to interrogate the damage, the baseline signal measured in absence of the damages; and

storing the received signal and deviation of the received signal from the baseline signal.

33. The computer readable medium of claim 32, wherein the one or more sequences of instructions implement a remote processing method of Simple Object Access Protocol (SOAP) or Remote Procedure Call/extensible Markup Language (RPC-XML) for Internet Webservices.

34. A computer readable medium carrying one or more sequences of instructions for identifying Lamb wave modes and determining time of arrivals of the Lamb wave modes, wherein execution of one or more sequences of instructions by one or more processors causes the one or more processors to perform the steps of:

loading a set of sensor signal data, each sensor signal data comprising Lamb wave signals measured at an excitation frequency;

detrending each of the set of sensor signal data to remove non-stationary signal component;

removing an electrical noise due to a toneburst actuator signal by applying a masking window to the each detrended sensor signal data;

performing a transformation on the each noise-removed sensor signal data to obtain a time-frequency signal energy distribution;

generating a multi-bandwidth energy distribution on a time-frequency plane by accumulating entire set of time-frequency signal energy distributions;

extracting one or more ridges from the multi-bandwidth energy distribution; and

identifying Lamb wave modes and determining time of arrivals of the Lamb wave modes based on the extracted one or more ridges.

35. The computer readable medium of claim 34, wherein the one or more sequences of instructions implements a remote processing method of Simple Object Access Protocol (SOAP) or Remote Procedure Call/extensible Markup Language (RPC-XML) for Internet Webservices.

36. A computer readable medium carrying one or more sequences of instructions for generating structural condition index (SCI) datasets from a plurality of sensor signal datasets, wherein execution of one or more sequences of instructions by one or more processors causes the one or more processors to perform the steps of:

(a) loading a plurality of sensor signal datasets for a plurality of network paths, each of the plurality of sensor signal datasets measured at one excitation frequency;

(b) selecting one of the plurality of sensor signal datasets;

(c) selecting one sensor signal from the selected sensor signal dataset, wherein the selected sensor signal is a Lamb wave signal;

(d) detrend and partition the selected sensor signal by applying an average filter and a masking window, respectively;

(e) decomposing the partitioned sensor signal into sub-bandwidth wave packets by applying a wavelet decomposition filter;

(d) synthesizing new sub-bandwidth packets;

(g) extracting S₀, S<sub>0—

ref</sub>, and A₀ modes from the synthesized sub-bandwidth packets by applying a set of envelop windows;

(h) computing at least one parameter of the set of envelop windows;

(i) determining a structural condition index for the selected sensor signal;

(j) determining a first discrete probability distribution function (DPDF) of the selected sensor signal;

(k) calculating a normality constant of the selected sensor signal;

(l) repeating the steps (d)-(k) for each sensor signal of the selected sensor signal dataset;

(m) determining a second DPDF for SCI dataset comprising structural condition indices obtained at step (i);

(n) finding and removing one or more outliers of the second DPDF;

(o) compensating an effect of ambient temperature on the SCI dataset;

(p) storing the SCI dataset; and

(q) repeating the steps (c)-(p) for each of the plurality of sensor signal datasets.

37. The computer readable medium of claim 36, wherein the one or more sequences of instructions implements a remote processing method of Simple Object Access Protocol (SOAP) or Remote Procedure Call/eXtensible Markup Language (RPC-XML) for Internet Webservices.

38. A computer readable medium carrying one or more sequences of instructions for generating a tomographic image to identify a region having changes in structural conditions that include damages of a host structure, wherein execution of one or more sequences of instructions by one or more processors causes the one or more processors to perform the steps of:

(a) loading a coordinate data of a plurality of diagnostic patches and a set of structural condition index (SCI) values for network paths defined by the plurality of diagnostic patches, the set of SCI values measured at an excitation frequency;

(b) calculating a bisection point for each of the network paths and assigning a corresponding one of the set of SCI values to the bisection point;

(c) calculating intersection points of the network paths;

(d) designating a SCI product to each of the intersection points;

(e) calculating SCI values near the intersection points using 3-dimensional SCI Gaussian functions, each of the 3-dimensional Gaussian functions defined for each of the network paths;

(f) generating a SCI distribution over a network plane by interpolation and a set of mesh-grid points of the network plane;

(g) setting up a chromosome population by assigning each chromosome to a corresponding one of the mesh-grid points;

(h) evaluating and ranking the chromosome population;

(i) selecting parent chromosomes from the chromosome population and reproducing child chromosomes;

(j) replacing the parent chromosomes with the reproduced child chromosomes;

(k) repeating the steps (i)-(j) over a preset number of generations to generate a final population of chromosomes;

(l) refining the SCI distribution over the final population of chromosomes; and

(m) generating a tomographic image of the refined SCI distribution.

39. The computer readable medium of claim 38, wherein the one or more sequences of instructions implements a remote processing method of Simple Object Access Protocol (SOAP) or Remote Procedure Call/extensible Markup Language (RPC-XML) for Internet Webservices.

40. A computer readable medium carrying one or more sequences of instructions for generating a tomographic image to identify changes in structural conditions or damages of a host structure, wherein execution of one or more sequences of instructions by one or more processors causes the one or more processors to perform the steps of:

loading a time of arrival dataset of a Lamb wave mode for a plurality of network paths defined by a plurality of diagnostic patches, the plurality of diagnostic patches applied to the host structure;

applying an algebraic reconstruction technique to reconstruct the loaded time of arrival dataset; and

generating a tomographic image of entire region of the host structure based on the reconstructed dataset.

41. The computer readable medium of claim 40, wherein the one or more sequences of instructions implements a remote processing method of Simple Object Access Protocol (SOAP) or Remote Procedure Call/extensible Markup Language (RPC-XML) for Internet Webservices.