Buried object locating and tracing method and system employing principal components analysis for blind signal detection

• US 7,136,765 B2
• Filed: 08/15/2005
• Issued: 11/14/2006
• Est. Priority Date: 02/09/2005
• Status: Active Grant

• Current Assignee: Seek Tech, Inc.
• Original Assignee: Deep Sea Power & Light Incorporated
• Inventors: Merewether, Ray, Olsson, Mark S., Maier, Christoph H.
• Primary Examiner(s): Tsai, Carol S. W.
• Application Number:
  US11/205,267
• Publication Number:
  US 20060178849A1
• Time in Patent Office: 456 Days
• Field of Search: None
• US Class Current: 702/65
• CPC Class Codes: G01V 3/12 operating with electromagnetic waves operating with millimetre waves G01V8/005

Abstract

A blind locating system for finding and tracing buried objects such as utility lines, conductive pipes and sondes. A sensor array is coupled to a signal processor, which determines a field vector for one or more buried objects by producing a data signal representing a covariance matrix corresponding to the covariances of the time-varying sensor array signals over a selected frequency band and accumulation interval. The covariance matrix is characterized by eigenvalues and associated eigenvectors and a user interface (UI) indicates the field vector associated with the eigenvector having the largest eigenvalue. Using several different frequency bands, a plurality of underground objects may be simultaneously detected and indicated in the UI without foreknowledge of their existence or characteristics.

Claims

1. An apparatus for locating one or more buried objects each characterized by an electromagnetic signal emission, the apparatus comprising:

a sensor assembly including a plurality K of sensors each having a sensor axis and producing a time-varying sensor signal S_k(t), wherein 1≦

k≦

K;

a processor coupled to the sensor assembly for processing a plurality K of the time-varying sensor signals {S_k(t)} representing a first electromagnetic signal emission, including

a matrix accumulator for producing a data signal representing a K by K covariance matrix A_T corresponding to the covariances of the time-varying sensor signals {S_k(t)} over a selected time interval, wherein covariance matrix A_T is characterized by a plurality K of eigenvalues {λ

_k} and associated eigenvectors {V_k} and

calculating means for producing a data signal representing a field vector of the first electromagnetic signal emission at the sensor assembly, wherein the field vector corresponds to the eigenvector V₁ associated with the largest λ

₁ of the eigenvalues {λ

_k}; and

a user interface (UI) coupled to the processing circuit for indicating the first electromagnetic signal emission field vector at the sensor assembly.

2. The apparatus of claim 1 further comprising:

a bandpass filter circuit coupled between the sensor assembly and the processor for blocking the portions of the time-varying sensor signals {S_k(t)} outside of a predetermined frequency region and passing a plurality K of band-limited sensor signals {S′

_k(t)} to the processor.

3. The apparatus of claim 2 further comprising:

a plurality F of the bandpass filter circuits each coupled between the sensor assembly and the processor for producing a plurality K of band-limited sensor signals {S′

_k(t)}_f for each of a plurality F of predetermined frequency regions; and

a plurality of the processors each for processing an f^th plurality K of band-limited sensor signals {S′

_k(t)}_f representing one of a plurality E of the electromagnetic signal emissions, wherein 1≦

f≦

F,

thereby simultaneously producing data signals representing field vectors of the plurality E of electromagnetic signal emissions characterizing a plurality of buried objects.

4. The apparatus of claim 3 with the sensor assembly further comprising:

a first sensor array including three sensors whose sensor axes are disposed to substantially intersect a first array center point orthogonally to one another;

a second sensor array including one or more sensors whose sensor axes are disposed to substantially intersect a second array center point; and

an elongate member extending through the first and second array center points and supporting the first and second sensor arrays in a fixed spatial relationship.

5. The apparatus of claim 3 with the bandpass filter circuit includes at least one filter circuit selected from a group comprising:

an Infinite Impulse Response (IIR) filter circuit;

a Fast-Fourier-Transform (FFT) filter circuit; and

an Finite Impulse Response (FIR) filter circuit.

6. The apparatus of claim 3 with the UI further comprising:

means for simultaneously indicating a plurality E of electromagnetic signal emission field vectors at the sensor assembly.

7. The apparatus of claim 1 further comprising:

a high-pass filter coupled to the sensor assembly for producing a plurality K of de-meaned time-varying sensor signals {S″

_k(t)} representing the first electromagnetic signal emission; and

in the processor, a second matrix accumulator coupled to the high pass filter for producing a data signal representing a K by K correlation matrix C_T corresponding to the correlations of the de-meaned time-varying sensor signals {S″

_k(t)} over a selected time interval, wherein correlation matrix C_T is characterized by the plurality K of eigenvalues {λ

_k} and associated eigenvectors {V_k}.

8. The apparatus of claim 1 further comprising:

a pre-rotation processor in the sensor assembly for processing a plurality K of first time-varying sensor signals each aligned with the corresponding sensor axis, including

means for producing a data signal representing the sensor assembly attitude with respect to at least one dimension of a predetermined coordinate system, and

means for producing a data signal representing the plurality K of time-varying sensor signals {S_k(t)} corresponding to the plurality K of first time-varying sensor signals rotated with respect to the sensor assembly into alignment with the predetermined coordinate system.

9. The apparatus of claim 1 further comprising:

in the processor, a virtual subgroup processor for selecting a plurality J of the K sensors to form a plurality G first subgroups each providing a plurality J_g<

K of the sensor signals {S_j(t)}_g, where 1≦

g≦

G, including

means for processing one or more of the plurality G first subgroups to obtain a data signal representing a field vector of the first electromagnetic signal emission at each of the processed first subgroups; and

means for processing a second subgroup formed by at least one sensor from each of the first subgroups to normalize the field vectors across all of the processed subgroups.

10. The apparatus of claim 1 with the calculating means further comprising:

means for producing a data signal representing an azimuth angle θ

_AZ of the electromagnetic signal emission field vector at the sensor assembly.

11. The apparatus of claim 1 with the calculating means further comprising:

means for producing a data signal representing an elevation angle φ

_EL of the electromagnetic signal emission field vector at the sensor assembly.

12. The apparatus of claim 1 with the UI further comprising:

means for simultaneously indicating a plurality E of electromagnetic signal emission field vectors at the sensor assembly.

13. The apparatus of claim 1 with the sensor assembly further comprising:

a first sensor array including three sensors whose sensor axes are disposed to substantially intersect a first array center point orthogonally to one another;

a second sensor array including one or more sensors whose sensor axes are disposed to substantially intersect a second array center point; and

an elongate member extending through the first and second array center points and supporting the first and second sensor arrays in a fixed spatial relationship.

14. A machine-implemented method for locating an electromagnetic signal emission with respect to a plurality K of electromagnetic sensors each producing a time-varying sensor signal S_k(t), wherein 1≦

k≦

K, the method comprising the steps of;

(a) producing a data signal representing a K by K covariance matrix A_T corresponding to the covariances of the plurality K of time-varying sensor signals {S_k(t)} over a selected time interval, wherein covariance matrix A_T is characterized by a plurality K of eigenvalues {λ

_k} and associated eigenvectors {V_k}; and

(b) producing a data signal representing a field vector of the electromagnetic signal emission at the electromagnetic sensor plurality, from which the electromagnetic signal emission location may be inferred, wherein the field vector corresponds to the eigenvector V₁ associated with the largest λ

₁ of the eigenvalues {λ

_k}.

15. The method of claim 14 further comprising the steps of:

(a.1) producing a data signal representing a diagonal singular value decomposition matrix Σ

of the covariance matrix A_T such that A_T=UΣ

V^T and U and V are orthogonal matrices;

(a.2) producing a data signal representing the plurality K of eigenvalues {λ

_k} and associated eigenvectors {V_k} corresponding to the diagonal singular value decomposition matrix Σ

.

16. The method of claim 15 further comprising the steps of:

(b.1) producing a data signal representing an azimuth angle θ

_AZ of the electromagnetic signal emission field vector at the electromagnetic sensor plurality.

17. The method of claim 16 further comprising the steps of:

(b.1) producing a data signal representing an elevation angle φ

_EL of the electromagnetic signal emission field vector at the electromagnetic sensor plurality.

18. The method of claim 14 wherein a plurality J of the K sensors are selected to form a plurality G of first subgroups each providing a plurality J_g<

K of the sensor signals {S_j(t)}_g, where 1≦

g≦

G, further comprising the steps of;

(c.1) performing steps (a) and (b) to process one or more of the plurality G of first subgroups to obtain a data signal representing a field vector of the electromagnetic signal emission at each of the processed first subgroups; and

(c.2) performing steps (a) and (b) to process a second subgroup formed by at least one sensor from each of the first subgroups to normalize the field vectors across all of the processed subgroups.

19. The method of claim 14 further comprising the steps of:

(b.1) producing a data signal representing an azimuth angle θ

_AZ of the electromagnetic signal emission field vector at the electromagnetic sensor plurality.

20. A machine-implemented method for simultaneously locating a plurality E of uncorrelated electromagnetic signal emissions with respect to a plurality K of electromagnetic sensors each producing a time-varying sensor signal S_k(t), wherein 1≦

k≦

K, the method comprising the steps of;

(a) filtering the plurality K of time-varying sensor signals {S_k(t)} to obtain a plurality K of band-limited sensor signals {S′

_k(t)}_f for each of a plurality F of predetermined frequency regions;

wherein 1≦

f≦

F; and

(b) processing each of the plurality F of band-limited sensor signal pluralities {S′

_k(t)}_f by performing the steps of

(b.1) producing a data signal representing a K by K covariance matrix A_Tf corresponding to the covariances of the f^th plurality K of time-varying sensor signals {S<sub>′

k</sub>(t)}_f over a selected time interval, wherein covariance matrix A_Tf is characterized by a plurality K of eigenvalues {λ

_k}_f and associated eigenvectors {V_k}_f and

(b.2) producing a data signal representing a field vector of the f^th electromagnetic signal emission at the electromagnetic sensor plurality, from which at least one uncorrelated electromagnetic signal emission location may be inferred, wherein the field vector corresponds to the eigenvector V_1f associated with the largest λ

_1f of the eigenvalues {λ

_k}_f.

21. The method of claim 20 further comprising the steps of:

(b.1.1) producing a data signal representing a diagonal singular value decomposition matrix Σ

_f of the f^th covariance matrix A_Tf such that A_Tf=U_fΣ

_fV_f^T and U_f and V_f are orthogonal matrices;

(b.1.2) producing a data signal representing the plurality K of eigenvalues {λ

_k}_f and associated eigenvectors {V_k}_f corresponding to the diagonal singular value decomposition matrix Σ

_f.

22. The method of claim 21 further comprising the steps of:

(b.2.1) producing a data signal representing an azimuth angle θ

_AZ of the electromagnetic signal emission field vector at the electromagnetic sensor plurality.

23. The method of claim 21 further comprising the steps of:

(b.2.1) producing a data signal representing an elevation angle φ

_EL of the electromagnetic signal emission field vector at the electromagnetic sensor plurality.

24. The method of claim 20 wherein a plurality J of the K sensors are selected to form a plurality G first subgroups each providing a plurality J_g<

K of sensor signals {S_j(t)}_g, where 1≦

g≦

G, further comprising the steps of;

(c.1) performing steps (a) and (b) to process one or more of the plurality G first subgroups to obtain a data signal representing a field vector of the electromagnetic signal emission at each of the processed first subgroups; and

(c.2) performing steps (a) and (b) to process a second subgroup formed by at least one member from each of the first subgroups to normalize the field vectors across all of the processed subgroups.

25. The method of claim 20 further comprising the steps of:

(b.2.1) producing a data signal representing an azimuth angle θ

_AZ of the electromagnetic signal emission field vector at the electromagnetic sensor plurality.

26. A machine-implemented method for locating an electromagnetic signal emission with respect to a plurality K of electromagnetic sensors each producing a de-meaned time-varying sensor signal S_k(t), wherein 1≦

k≦

K, the method comprising the steps of;

(a) producing a data signal representing a K by K correlation matrix C_T corresponding to the correlations of the plurality K of de-meaned time-varying sensor signals {S_k(t)} over a selected time interval, wherein correlation matrix C_T is characterized by a plurality K of eigenvalues {λ

_k} and associated eigenvectors {V_k}; and

(b) producing a data signal representing a field vector of the electromagnetic signal emission at the electromagnetic sensor plurality from which the electromagnetic signal emission location may be inferred, wherein the field vector corresponds to the eigenvector V₁ associated with the largest λ

₁ of the eigenvalues {λ

_k}.

27. The method of claim 26 further comprising the steps of:

(a.1) producing a data signal representing a diagonal singular value decomposition matrix Σ

of the correlation matrix C_T such that A_T=UΣ

V^T and U and V are orthogonal matrices;

(a.2) producing a data signal representing the plurality K of eigenvalues {λ

_k} and associated eigenvectors {V_k} corresponding to the diagonal singular value decomposition matrix Σ

.

28. The method of claim 27 further comprising the steps of:

(b.1) producing a data signal representing an azimuth angle θ

_AZ of the electromagnetic signal emission field vector at the electromagnetic sensor plurality.

29. The method of claim 28 further comprising the steps of:

(b.1) producing a data signal representing an elevation angle φ

_EL of the electromagnetic signal emission field vector at the electromagnetic sensor plurality.

30. The method of claim 26 wherein a plurality J of the K sensors are selected to form a plurality G of first subgroups each providing a plurality J_g<

K of the sensor signals {S_j(t)}_g, where 1≦

g≦

G, further comprising the steps of;

(c.1) performing steps (a) and (b) to process one or more of the plurality G of first subgroups to obtain a data signal representing a field vector of the first electromagnetic signal emission at each of the processed first subgroups; and

(c.2) performing steps (a) and (b) to process a second subgroup formed by at least one sensor from each of the first subgroups to normalize the field vectors across all of the processed subgroups.

31. The method of claim 26 further comprising the steps of:

(b.1) producing a data signal representing an azimuth angle θ

_AZ of the electromagnetic signal emission field vector at the electromagnetic sensor plurality.

32. A machine-implemented method for locating an electromagnetic signal emission with respect to a plurality K of electromagnetic sensors each having a sensor axis oriented with respect to at least one dimension of a predetermined coordinate system and producing a time-varying sensor signal B_k(t), wherein 1≦

k≦

K, the method comprising the steps of;

(a) producing a plurality K of time-varying sensor signals {S_k(t)} representing a plurality K of time-varying sensor signals {B_k(t)} rotated into alignment with the at least one predetermined coordinate system dimension;

(b) producing a data signal representing a K by K covariance matrix A_T corresponding to the covariances of the plurality K of time-varying sensor signals {S_k(t)} over a selected time interval, wherein covariance matrix A_T is characterized by a plurality K of eigenvalues {λ

_k} and associated eigenvectors {V_k}; and

(c) producing a data signal representing a field vector of the electromagnetic signal emission at the electromagnetic sensor plurality from which the electromagnetic signal emission location may be inferred, wherein the field vector corresponds to the eigenvector V₁ associated with the largest λ

₁ of the eigenvalues {λ

_k}.

33. The method of claim 32 further comprising the steps of:

(a.1) producing a data signal representing a diagonal singular value decomposition matrix Σ

of the covariance matrix A_T such that A_T=UΣ

V^T and U and V are orthogonal matrices;

(a.2) producing a data signal representing the plurality K of eigenvalues {λ

_k} and associated eigenvectors {V_k} corresponding to the diagonal singular value decomposition matrix Σ

.

34. The method of claim 32 wherein a plurality J of the K sensors are selected to form a plurality G of first subgroups each having a plurality J_g<

K of the sensor signals {S_j(t)}_g, where 1≦

g≦

G, further comprising the steps of;

(c.1) performing steps (a) and (b) to process one or more of the plurality G of first subgroups to obtain a data signal representing a field vector of the electromagnetic signal emission at each of the processed first subgroups; and

(c.2) performing steps (a) and (b) to process a second subgroup formed by at least one member from each of the first subgroups to normalize the field vectors across all of the processed subgroups.