SIGNAL PROCESSING FAULT DETECTION SYSTEM

• US 20090016597A1
• Filed: 07/13/2007
• Published: 01/15/2009
• Est. Priority Date: 07/13/2007
• Status: Active Grant

• Current Assignee: United States Of America Naval Undersea Warfare Center
• Original Assignee: United States Of America Naval Undersea Warfare Center
• Inventors: HUGHES, DERKE R., Katz, Richard A., Nuttall, Albert H.
• Granted Patent: US 7,620,503 B2
• Time in Patent Office: Days
• US Class Current: 382/149
• 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

Abstract

A fault detection system designed to evaluate the structural integrity of a material employs an array of sensors disposed over the material being evaluated. The sensors detect vibrations in the material and the sensor signals are fed to a data processor. The processor employs a method to analyze the linear and nonlinear characteristics of the sensor signals and then determines whether to proceed with a linear signal processing analysis or a nonlinear signal processing analysis of the sensor signals. Once the analysis is completed, the results are compared to baseline results to determine what if any divergence exists between the results and the baseline results. A significant divergence indicates a potential material failure. The fault detection system will indicate such a potential failure through a visual alarm on a graphical user interface.

Claims

1. A method for detecting material faults such as cracks in a structure comprising:

modeling said structure on a computer as a structure with no significant material faults;

obtaining time series data from the computer modeled structure wherein said time series data represents vibrations in the computer modeled structure;

performing a nonlinear analysis on the time series data obtained from the computer modeled structure to use as a baseline of values for said structure with no significant faults;

deploying a plurality of sensors capable of detecting vibrations on the structure in order to detect vibrations in the structure, wherein each of said plurality of sensors generates a sensor output signal in the form of time series data;

obtaining experimental time series data from said plurality of sensors;

performing a preliminary nonlinear analysis on the experimental time series data from said plurality of sensors;

determining whether said experimental time series data from said plurality of sensors is nonlinear based on said preliminary nonlinear analysis;

determining whether said experimental time series data from said plurality of sensors is linear based on said preliminary nonlinear analysis;

performing a linear analysis on said experimental time series data from said plurality of sensors based on the determination that the data was linear;

performing a further nonlinear analysis on said experimental time series data from said plurality of sensors based on the determination that the data was nonlinear; and

determining a damage level of the structure based on the variance between the experimental time series data and said baseline of values for said structure with no significant faults, wherein said damage level is related to a specific tolerance for said structure.

2. The method of claim 1 further comprising the step of indicating visually the damage level of the structure on a graphical user interface.

3. The method of claim 1 wherein the step of performing a non linear analysis on the time series data obtained from the computer modeled structure to use as a base line comprises performing a mutual information analysis according to the following equation:

4. The method of claim 1 wherein said sensors capable of detecting vibrations on the structure are piezoelectric strain sensors.

5. The method of claim 1 wherein the step of performing a preliminary non linear analysis on the time series data obtained from the plurality of sensors comprises performing a mutual information analysis according to the following equation:

6. The method of claim 1 wherein the step of performing a linear analysis on said experimental time series data from said plurality of sensors comprises performing a wavelet transform function analysis according to the following equation:

7. The method of claim 1 wherein the step of performing a linear analysis on said experimental time series data from said plurality of sensors comprises performing a fast Fourier transform analysis according to the following equation:

8. The method of claim 1 wherein the step of performing a further nonlinear analysis on said experimental time series data from said plurality of sensors comprises performing a mutual information analysis combined with a surrogate data method analysis according to the following equations:

16. The method of claim 1 wherein the step of performing a further nonlinear analysis on said experimental time series data from said second plurality of sensors comprises performing a mutual information analysis combined with a surrogate data method analysis according to the following equations:

9. A method for detecting material faults such as cracks in a structure comprising:

deploying a first plurality of sensors capable of detecting vibrations on a first structure with no significant material faults in order to detect vibrations in said first structure, wherein each of said first plurality of sensors generates a sensor output signal in the form of time series data;

obtaining time series data from the first plurality of sensors structure wherein said time series data represents vibrations in the first structure;

performing a nonlinear analysis on the time series data obtained from the first plurality of sensors to use as a baseline of values for said first structure with no significant faults;

deploying a second plurality of sensors capable of detecting vibrations on a second structure in order to detect vibrations in said second structure, wherein each of said plurality of sensors generates a sensor output signal in the form of time series data;

obtaining experimental time series data from said second plurality of sensors;

performing a preliminary nonlinear analysis on the experimental time series data from said second plurality of sensors;

determining whether said experimental time series data from said second plurality of sensors is nonlinear based on said preliminary nonlinear analysis;

determining whether said experimental time series data from said second plurality of sensors is linear based on said preliminary nonlinear analysis;

performing a linear analysis on said experimental time series data from said second plurality of sensors based on the determination that the data was linear;

performing a further nonlinear analysis on said experimental time series data from said second plurality of sensors based on the determination that the data was nonlinear; and

determining a damage level of the second structure based on the variance between the experimental time series data and said baseline of values for said first structure with no significant faults, wherein said damage level is related to a specific tolerance for said first structure.

10. The method of claim 9 further comprising the step of indicating visually the damage level of said second structure on a graphical user interface.

11. The method of claim 9 wherein the step of performing a non linear analysis on the time series data obtained from the first plurality of sensors to use as a base line comprises performing a mutual information analysis according to the following equation:

12. The method of claim 9 wherein said sensors capable of detecting vibrations on a structure are piezoelectric strain sensors.

13. The method of claim 9 wherein the step of performing a preliminary non linear analysis on the time series data obtained from the second plurality of sensors comprises performing a mutual information analysis according to the following equation:

14. The method of claim 9 wherein the step of performing a linear analysis on said experimental time series data from said second plurality of sensors comprises performing a wavelet transform function analysis according to the following equation:

15. The method of claim 9 wherein the step of performing a linear analysis on said experimental time series data from said second plurality of sensors comprises performing a fast Fourier transform analysis according to the following equation: