Method and apparatus for categorizing unexplained residuals

• US 6,961,677 B1
• Filed: 08/25/2003
• Issued: 11/01/2005
• Est. Priority Date: 08/25/2003
• Status: Active Grant

• Current Assignee: Harris Corporation (L3Harris Technologies, Inc.)
• Original Assignee: ITT Manufacturing Enterprises, Inc. (ITT, Inc.)
• Inventors: Boysworth, Marc Kenneth
• Primary Examiner(s): Tsai, Carol S. W.
• Application Number:
  US10/646,774
• Time in Patent Office: 799 Days
• Field of Search: 702/84, 702/104, 702/183, 702189-191, 702/196, 700/29, 700/33, 700/44, 700/38, 708/300, 73/2, 73/7, 706/14, 706/24, 706/33, 706/52
• US Class Current: 702/189
• CPC Class Codes: G06F 18/00 Pattern recognition
• CPC Class Codes: G06F 2218/04 Denoising

Abstract

A method and apparatus for performing continuous variance analysis (CVA) to characterize a data set. Data set values may be associated with any source, including measurements of a received signal and/or measurements of natural and/or man-made phenomena. CVA generates an output matrix that contains a measure of variation for a plurality of ranges (or windows) of data elements within a data set positioned at known locations within the data set. CVA output can be interpreted visually by a technician and/or using automatic numerical analysis. CVA is compatible with any apparatus/approach that uses numerical analysis to generate a predicted model based upon stored library models and/or linear/nonlinear components. CVA is compatible with any programming language and can be readily added to new and/or existing apparatus to compliment existing capabilities. CVA is less complex than conventional techniques, and requires less computer processing capacity, yet results in more readily interpretable results.

Claims

1. A method of analyzing a set of values, comprising:

(a) defining a plurality of subsets of contiguous values within a set of values comprising residual values that represent a difference between data element values of a predicted model data set and data element values of a measured data set;

(b) determining a measure of variation among contiguous values in each of the plurality of subsets to produce a plurality of measures of variation corresponding to the plurality of subsets; and

(c) categorizing the set of values based upon an analysis of the plurality of measures of variation.

2. The method of claim 1, wherein the set of values are values measured from a communication signal.

3. The method of claim 1, wherein the set of values are values measured from one of an observed substance and an observed event.

4. The method of claim 1, wherein the residual values are the result of numerical analysis of a communication signal.

5. The method of claim 1, wherein the residual values are the result of numerical analysis of values associated with one of an observed substance and an observed event.

6. The method of claim 1, wherein (c) comprises characterizing the set of values as one of homoscedastic and heteroscedastic.

7. The method of claim 1, wherein (a) further comprises:

(a.1) defining a range of values not greater than a number of values within the set of values; and

(a.2) defining a subset of values by positioning the range at a specific position within the set of values.

8. The method of claim 7, wherein (a) further includes:

(a.3) varying the size of the range for a plurality of the subsets.

9. The method of claim 7, wherein (a) further includes:

(a.3) varying the position of the range within the set of values for a plurality of the subsets.

10. The method of claim 7, wherein (a) further includes:

(a.3) varying the size of the range for a plurality of the subsets; and

(a.4) varying the position of the range within the set of values for a plurality of the subsets.

11. The method of claim 1, wherein (b) further comprises:

(b.1) storing the determined measure of variation.

12. The method of claim 11, wherein (b.1) further comprises:

(b.1.1) storing a determined measure of variation in association with a size of the range and a position of the range associated with the subset for which the measure of variation was determined.

13. The method of claim 1, wherein (c) further comprises:

(c.1) categorizing the set of values based upon a difference between a measure of variation determined for one of the plurality of subsets and a measure of variation determined for another one of the plurality of subsets.

14. The method of claim 1, wherein (c) further comprises:

(c.1) categorizing the set of values based upon n-way principal component analysis of the measures of variation determined for the plurality of subsets.

15. The method of claim 1, wherein (c) further comprises:

(c.1) categorizing the set of values based upon visual analysis of a plot of the measures of variation determined for the plurality of subsets.

16. The method of claim 15, wherein the visual analysis is based upon patterns within one of a two-dimensional plot and a three-dimensional plot of the measures of variation determined for the plurality of subsets.

17. The method of claim 15, wherein a position of a measure of variation within the plot is based upon a size of the subset and a position of the subset for which the measure of variation was determined.

18. An apparatus for analyzing a set of values, comprising:

(a) a measurement module that produces a set of values comprising residual values that represent a difference between data element values of a predicted model data set and data element values of a measured data set;

(b) a windowing module that defines a plurality of subsets of contiguous values within the set of values;

(c) an analysis module that determines a measure of variation among contiguous values in each of the plurality of subsets to produce a plurality of measures of variation corresponding to the plurality of subsets; and

(d) an assessment module that categorizes the set of values based upon an analysis of the plurality of measures of variation.

19. The apparatus of claim 18, wherein the measurement module produces the measurement values measured from a communication signal.

20. The apparatus of claim 18, wherein the measurement module produces the measurement values measured from one of an observed substance and an observed event.

21. The apparatus of claim 18, wherein the measurement module produces the residual values as a result of numerical analysis of a communication signal.

22. The apparatus of claim 18, wherein the measurement module produces the residual values as a result of numerical analysis of values associated with one of an observed substance and an observed event.

23. The apparatus of claim 18, wherein the assessment module characterizes the set of values as one of homoscedastic and heteroscedastic.

24. The apparatus of claim 18, wherein the windowing module defines a range of values not greater than a number of values within the set of values.

25. The apparatus of claim 18, wherein the windowing module defines a subset of values by positioning the range at a specific position within the set of values.

26. The apparatus of claim 24, wherein the windowing module further varies the size of the range for a plurality of the subsets.

27. The apparatus of claim 25, wherein the windowing module varies the position of the range within the set of values for a plurality of the subsets.

28. The apparatus of claim 18, wherein the analysis module stores the determined measure of variation.

29. The apparatus of claim 28, wherein the analysis module stores a determined measure of variation in association with the size of the range and the position of the range associated with the subset for which the measure of variation was determined.

30. The apparatus of claim 18, wherein the assessment module categorizes the set of values based upon a difference between a measure of variation determined for one of the plurality of subsets and a measure of variation determined for another one of the plurality of subsets.

31. The apparatus of claim 18, wherein the assessment module categorizes the set of values based upon n-way principal component analysis of the measures of variation determined for the plurality of subsets.

32. The apparatus of claim 18, wherein the assessment module categorizes the set of values based upon visual analysis of a plot of the measures of variation determined for the plurality of subsets.

33. The apparatus of claim 32, wherein the visual analysis is based upon patterns within one of a two-dimensional plot and a three-dimensional plot of the measures of variation determined for the plurality of subsets.

34. The apparatus of claim 32, wherein a position of a measure of variation within the plot is based upon a size of the subset and a position of the subset for which the measure of variation was determined.

35. A program product apparatus having a computer readable medium with computer program logic recorded thereon for analyzing a set of values, said program product apparatus comprising:

(a) a measurement module that produces a set of values comprising residual values that represent a difference between data element values of a predicted model data set and data element values of a measured data set;

(b) a windowing module that defines a plurality of subsets of contiguous values within the set of values;

(c) an analysis module that determines a measure of variation among contiguous values in each of the plurality of subsets to produce a plurality of measures of variation corresponding to the plurality of subsets; and

(d) an assessment module that categorizes the set of values based upon an analysis of the plurality of measures of variation.

36. The program product apparatus of claim 35, wherein the assessment module characterizes the set of values as one of homoscedastic and heteroscedastic.

37. The program product apparatus of claim 35, wherein the assessment module categorizes the set of values based upon a difference between a measure of variation determined for one of the plurality of subsets and a measure of variation determined for another one of the plurality of subsets.

38. The program product apparatus of claim 35, wherein the assessment module categorizes the set of values based upon n-way principal component analysis of the measures of variation determined for the plurality of subsets.

39. The program product apparatus of claim 35, wherein the assessment module categorizes the set of values based upon visual analysis of a plot of the measures of variation determined for the plurality of subsets.

40. An apparatus for analyzing a set of values, comprising:

(a) means for producing a set of values comprising residual values that represent a difference between data element values of a predicted model data set and data element values of a measured data set;

(b) means for defining a plurality of subsets of contiguous values within the set of values;

(c) means for determining a measure of variation among contiguous values in each of the plurality of subsets to produce a plurality of measures of variation corresponding to the plurality of subsets; and

(d) means for categorizing the set of values based upon an analysis of the plurality of measures of variation.

41. The apparatus of claim 40, wherein said (c) means for categorizing characterizes the set of values as one of homoscedastic and heteroscedastic.

42. The apparatus of claim 40, wherein said (c) means for categorizing further comprises:

(c.1) means for categorizing the set of values based upon a difference between a measure of variation determined for one of the plurality of subsets and a measure of variation determined for another one of the plurality of subsets.