Full-field imaging learning machine (FILM)
A method of determining dynamic properties of a structure (linear or nonlinear) includes receiving spatio-temporal inputs, generating mode shapes and modal components corresponding to the spatio-temporal inputs using a trained deep complexity coding artificial neural network, and subsequently generating the dynamic properties by analyzing each modal component using a trained learning machine. A computing system for non-contact determination of dynamic properties of a structure includes a camera, a processor, and a memory including computer-executable instructions. When the instructions are executed, the system is caused to receive spatio-temporal image data, decompose the spatio-temporal image data into constituent manifold components using an autoencoder, and analyze the constituent manifold components using a trained learning machine to determine the dynamic properties.
1. A computer-implemented method of determining a set of nonlinear dynamic properties of a structure of interest, the method comprising:
receiving a set of spatio-temporal inputs,
generating a set of decomposed nonlinear modal components corresponding to the set of spatio-temporal inputs by analyzing the set of spatio-temporal inputs using a trained deep complexity coding artificial neural network, and
generating the set of nonlinear dynamic properties of the structure of interest by analyzing the set of decomposed nonlinear modal components using a trained learning machine.
2. The computer-implemented method of claim 1 , wherein the set of spatio-temporal inputs includes RGB-D, infrared, hyperspectral, microscopic, satellite, and/or other types of video data.
3. The computer-implemented method of claim 1 , wherein the set of spatio-temporal inputs correspond to each pixel in an image at a time step t.
4. The computer-implemented method of claim 1 , further comprising:
generating a dictionary including a full-field, pixel-level spatial characterization of the structure of interest.
5. The computer-implemented method of claim 4 , further comprising:
determining a health of the structure of interest based on the full-field, pixel-level spatial characterization of the structure of interest.
6. The computer-implemented method of claim 1 , further comprising:
determining, based on the set of decomposed nonlinear modal components, a temporal/spectral characterization.
7. The computer-implemented method of claim 6 , further comprising:
using the temporal/spectral characterization for a fatigue analysis, an aerodynamics analysis, an engine/rotor vibration analysis, or a material characterization.
8. The computer-implemented method of claim 1 , wherein the trained learning machine is a long short-term memory recurrent neural network.
9. The computer-implemented method of claim 1 , wherein the spatio-temporal input corresponds to video image data.
10. A computing system for non-contact determination of a set of nonlinear dynamic properties of a structure of interest, comprising:
a camera,
one or more processors,
a memory including computer-executable instructions that, when executed, cause the computing system to:
receive, via the camera, spatio-temporal image data,
decompose the spatio-temporal image data into a set of decomposed nonlinear modal components characterizing the full-field pixel-level structural properties of the structure of interest, by analyzing the spatio-temporal image data using an autoencoder, and
analyze the set of decomposed nonlinear modal components using a trained learning machine to determine the set of nonlinear dynamic properties.
11. The computing system of claim 10 , wherein the camera is a digital camera, digital video camera, a microscope, a scanning electronic microscope, an endoscope, a sonar device, or an infrared thermography camera.
12. The computing system of claim 10 , the memory including further computer-executable instructions that, when executed, cause the computing system to:
generate a dictionary including a full-field, pixel-level spatial characterization of the structure of interest.
13. The computing system of claim 10 , the memory including further computer-executable instructions that, when executed, cause the computing system to:
determine a health of the structure of interest based on the full-field, pixel-level spatial characterization of the structure of interest.
14. The computing system of claim 10 , the memory including further computer-executable instructions that, when executed, cause the computing system to:
determine, based on the set of decomposed nonlinear modal components, a temporal/spectral characterization.
15. The computing system of claim 10 , wherein the trained learning machine is a long short-term memory recurrent neural network.
16. A non-transitory computer readable medium containing program instructions that when executed, cause a computer to:
receive, via a camera, spatio-temporal image data,
decompose the spatio-temporal image data into a set of nonlinear modal components characterizing the full-field pixel-level structural properties of the structure of interest, by analyzing the spatio-temporal image data using an autoencoder, and
analyze the set of decomposed nonlinear modal components using a trained learning machine to determine the set of nonlinear dynamic properties.
17. The non-transitory computer readable medium of claim 16 , containing further program instructions that when executed, cause a computer to:
generate a dictionary including a full-field, pixel-level spatial characterization of the structure of interest.
18. The non-transitory computer readable medium of claim 16 , containing further program instructions that when executed, cause a computer to:
determine a health of the structure of interest based on the full-field, pixel-level spatial characterization of the structure of interest.
19. The non-transitory computer readable medium of claim 16 , containing further program instructions that when executed, cause a computer to:
determine, based on the set of decomposed nonlinear modal components, a temporal/spectral characterization.
20. The non-transitory computer readable medium of claim 16 , wherein the trained learning machine is a long short-term memory recurrent neural network.