Integrated method for chaotic time series analysis

• US 5,815,413 A
• Filed: 05/08/1997
• Issued: 09/29/1998
• Est. Priority Date: 05/08/1997
• Status: Expired due to Term

• Current Assignee: Lockheed Martin Energy Research Corporation (Martin Marietta Corporation)
• Original Assignee: Lockheed Martin Energy Research Corporation (Martin Marietta Corporation)
• Inventors: Ng, Esmond G., Hively, Lee M.
• Primary Examiner(s): Trammel, James P.
• Assistant Examiner(s): Bui, Bryan
• Application Number:
  US08/853,226
• Time in Patent Office: 509 Days
• Field of Search: 364/572, 364/573, 364/574, 364/280, 364/283.1, 364/286, 364/286.2, 364/551.01, 364/940, 364/948.2, 707/102, 707/200, 707/500, 707/530, 600/544, 600/545, 600/546, 126/702, 126/600, 126/621
• US Class Current: 702/191
• CPC Class Codes: G06F 17/18 for evaluating statistical data , e.g. average values, frequency distributions, probability functions, regression analysis forecasting specially adapted for a specific administrative, business or logistic context G06Q10/04
• CPC Class Codes: G06F 2218/12 Classification; Matching

Abstract

Methods and apparatus for automatically detecting differences between similar but different states in a nonlinear process monitor nonlinear data. Steps include: acquiring the data; digitizing the data; obtaining nonlinear measures of the data via chaotic time series analysis; obtaining time serial trends in the nonlinear measures; and determining by comparison whether differences between similar but different states are indicated.

Claims

1. A method for automatically discriminating between similar but different states in a nonlinear process comprising the steps of:

(A) Operating a data provision means selected from the group consisting of data storage means and data acquisition means to provide at least one channel of nonlinear data, called e-data;

(B) Separating the e-data into artifact data, called f-data, and artifact-free data, called g-data, while preventing phase distortions in the data;

(C) Processing g-data through a filter to produce a filtered version of g-data, called h-data;

(D) Applying at least one nonlinear measure to at least one type of data selected from the group consisting of e-data, f-data, g-data, and h-data to provide at least one time serial sequence of nonlinear measures from which at least one indicative trend selected from the group consisting of abrupt increases and abrupt decreases can be determined;

(E) Comparing at least one indicative trend with at least one known discriminating indicator;

(F) Determining from said comparison whether differences between similar but different states are indicated; and

(G) Providing notification whether differences between similar but different states are indicated,

Said steps B, C, D, E, and F being accomplishable in one integrated sequence of computer analyses.

2. The method as described in claim 1 wherein said at least one time serial sequence of nonlinear measures is selected from the group consisting of:

time per wave cycle for e-data, time per wave cycle for f-data, time per wave cycle for g-data, time per wave cycle for h-data, Kolmogorov entropy for e-data, Kolmogorov entropy for f-data, Kolmogorov entropy for g-data, Kolmogorov entropy for h-data, first minimum in the mutual information function for e-data, first minimum in the mutual information function for f-data, first minimum in the mutual information function for g-data, first minimum in the mutual information function for h-data, correlation dimension for e-data, correlation dimension for f-data, correlation dimension for g-data, correlation dimension for h-data, and combinations thereof.

3. The method as described in claim 1 wherein the e-data is separated into f-data and g-data by use of a zero-phase filter.

4. The method as described in claim 1 wherein the filter comprises a standard low-pass filter selected from the group consisting of second-order, third-order, and fourth-order low-pass filters at frequencies between about 35 Hz and about 60 Hz.

5. The method as described in claim 4 wherein the low-pass filter is a standard fourth-order low-pass filter at about 50 Hz.

6. Apparatus for automatically discriminating between similar but different states in a nonlinear process comprising:

(A) Data provision means for providing at least one channel of nonlinear data, called e-data, said data provision means being selected from the group consisting of data storage means and data acquisition means;

(B) Separation means for separating the e-data into artifact data, called f-data, and artifact-free data, called g-data, while preventing phase distortions in the data, communicably connected to said data provision means;

(C) Filter means for filtering g-data to produce a filtered version of g-data, called h-data, communicably connected to said separation means;

(D) Application means for applying at least one nonlinear measure to at least one type of data selected from the group consisting of e-data, f-data, g-data, and h-data to provide at least one time serial sequence of nonlinear measures, from which at least one indicative trend selected from the group consisting of abrupt increases and abrupt decreases can be determined, communicably connected to said filter means;

(E) Comparison means for comparing at least one indicative trend with at least one known discriminating indicator, communicably connected to said application means;

(F) Determination means for determining from said comparison whether differences between similar but different states are indicated, communicably connected to said comparison means; and

(G) Notification means for providing notification whether differences between similar but different states are indicated, communicably connected to said determination means,

Said elements B, C, D, E, and F being operable in one integrated sequence of computer analyses.

7. The apparatus as described in claim 6 wherein said at least one time serial sequence of nonlinear measures is selected from the group consisting of:

time per wave cycle for e-data, time per wave cycle for f-data, time per wave cycle for g-data, time per wave cycle for h-data, Kolmogorov entropy for e-data, Kolmogorov entropy for f-data, Kolmogorov entropy for g-data, Kolmogorov entropy for h-data, first minimum in the mutual information function for e-data, first minimum in the mutual information function for f-data, first minimum in the mutual information function for g-data, first minimum in the mutual information function for h-data, correlation dimension for e-data, correlation dimension for f-data, correlation dimension for g-data, correlation dimension for h-data, and combinations thereof.

8. The apparatus as described in claim 6 wherein the e-data is separated into f-data and g-data by use of a zero-phase filter.

9. The apparatus as described in claim 8 wherein said filter means comprises a standard low-pass filter selected from the group consisting of second-order, third-order, and fourth-order low-pass filters at frequencies between about 35 Hz and about 60 Hz.

10. The apparatus as described in claim 9 wherein said low-pass filter is a standard fourth-order low-pass filter at about 50 Hz.

11. A method for automatically discriminating between similar but different states in a nonlinear process comprising the steps of:

(A) Operating a data provision means selected from the group consisting of data storage means and data acquisition means to provide at least one channel of nonlinear data, called e-data;

(B) Separating the e-data into artifact data, called f-data, and artifact-free data, called g-data, while preventing phase distortions in the data;

(C) Processing g-data through a filter to produce a filtered version of g-data, called h-data;

(D) Applying the lag derived from the first minimum in the mutual information function to create a d-dimensional probability density function which forms a high-dimensional topology for at least one type of data selected from the group consisting of e-data, f-data, g-data, and h-data whereby at least one indicative trend can be determined;

(E) Comparing at least one indicative trend with at least one known discriminating indicator;

(F) Determining from said comparison whether differences between similar but different states are indicated; and

(G) Providing notification whether differences between similar but different states are indicated,

Steps B, C, D, E, and F being accomplishable in one integrated sequence of computer analyses.

12. The method as described in claim 11 wherein the e-data is separated into f-data and g-data by use of a zero-phase filter.

13. The method as described in claim 12 wherein the filter comprises a standard low-pass filter selected from the group consisting of second-order, third-order, and fourth-order low-pass filters at frequencies between about 35 Hz and about 60 Hz.

14. The method as described in claim 13 wherein the low-pass filter is a standard fourth-order low-pass filter at about 50 Hz.

15. A method for automatically discriminating between similar but different states in a nonlinear process comprising the steps of:

(A) Operating a data provision means selected from the group consisting of data storage means and data acquisition means to provide at least one channel of nonlinear data, called e-data;

(B) Separating the e-data into artifact data, called f-data, and artifact-free data, called g-data, while preventing phase distortions in the data;

(C) Processing g-data through a filter to produce a filtered version of g-data, called h-data;

(D) Applying the lag derived from the first minimum in the mutual information function to create a d-dimensional probability density function which forms a high-dimensional topology for at least one type of data selected from the group consisting of e-data, f-data, g-data, and h-data whereby at least one indicative trend can be determined;

(E) Generating a key from the set of indices of each bin in the multidimensional probability density function;

(F) Representing occupied bins from the same class by a linked list to provide a reduction in the number of elements to be stored in the manner known as hashing whereby at least one hashed discriminating trend can be determined;

(G) Comparing at least one indicative trend with at least one known discriminating indicator;

(H) Determining from said comparison whether differences between similar but different states are indicated; and

(I) Providing notification whether differences between similar but different states are indicated,

Said discriminating indicator being hashed, and Steps B, C, D, E, F, G, and H being accomplishable in one integrated sequence of computer analyses.

16. An apparatus for automatically discriminating between similar but different states in a nonlinear process comprising:

(A) Data provision means for providing at least one channel of nonlinear data, called e-data, said data provision means being selected from the group consisting of data storage means and data acquisition means;

(B) Separation means for separating the e-data into artifact data, called f-data, and artifact-free data, called g-data, while preventing phase distortions in the data, communicably connected to said data provision means;

(C) Filter means for filtering g-data to produce a filtered version of g-data, called h-data, communicably connected to said separation means;

(D) Application means for applying the lag derived from the first minimum in the mutual information function to create a d-dimensional probability density function which forms a high-dimensional topology for at least one type of data selected from the group consisting of e-data, f-data, g-data, and h-data whereby at least one indicative trend can be determined, communicably connected to said filter means;

(E) Comparison means for comparing at least one indicative trend with at least one known discriminating indicator, communicably connected to said application means;

(F) Determination means for determining from the comparison whether differences between similar but different states are indicated, communicably connected to said comparison means; and

(G) Notification means for providing notification whether differences between similar but different states are indicated, communicably connected to said determination means,

Said discriminating indicator being hashed and said elements B, C, D, E, and F being operable in one integrated sequence of computer analyses.

17. The apparatus as described in claim 16 wherein the e-data is separated into f-data and g-data by use of a zero-phase filter.

18. The apparatus as described in claim 17 wherein the filter means comprises a standard low-pass filter selected from the group consisting of second-order, third-order, and fourth-order low-pass filters at frequencies between about 35 Hz and about 60 Hz.

19. The apparatus as described in claim 18 wherein the low-pass filter is a standard fourth-order low-pass filter at about 50 Hz.

20. An apparatus for automatically discriminating between similar but different states in a nonlinear process comprising:

(A) Data provision means for providing at least one channel of nonlinear data, called e-data, said data provision means being selected from the group consisting of data storage means and data acquisition means;

(B) Separation means for separating the e-data into artifact data, called f-data, and artifact-free data, called g-data, while preventing phase distortions in the data, communicably connected to said data provision means;

(C) Filter means for filtering g-data to produce a filtered version of g-data, called h-data, communicably connected to said separation means;

(D) Application means for applying the lag derived from the first minimum in the mutual information function to create a d-dimensional probability density function which forms a high-dimensional topology for at least one type of data selected from the group consisting of e-data, f-data, g-data, and h-data whereby at least one indicative trend can be determined, communicably connected to said filter means;

(E) Generation means for generating a key from the set of indices of each bin in the multidimensional probability density function, communicably connected to said application means;

(F) Representation means for representing occupied bins from the same class by a linked list to provide a reduction in the number of elements to be stored in the manner known as hashing, whereby at least one hashed indicative trend can be determined, communicably connected to said generation means;

(G) Comparison means for comparing at least one indicative trend with at least one known discriminating indicator, communicably connected to said representation means;

(H) Determination means for determining from said comparison whether differences between similar but different states are indicated, communicably connected to said comparison means; and

(I) Notification means for providing notification whether differences between similar but different states are indicated, communicably connected to said determination means,

Said discriminating indicator being hashed, and Elements B, C, D, E, F, G, and H being operable in one integrated sequence of computer analyses.