Method and system for analysis of ultrasonic reflections in real time
View Patent ↗An ultrasonic real-time inspection method which is user-friendly in an interactive environment to provide ease of operation, as well as a combination of consistency, thoroughness, and speed of operation in flaw detection not achievable by other methods. The method offers significantly increased pattern recognition capability, which provides greater automation potential and reduced missed detection and false alarm rates.
1. A method of performing real-time pattern recognition on an elongated material, said method comprising the steps of:
injecting, using a test device, pulses of a plurality of ultrasonic beams into the elongated material;
detecting acoustic responses caused at a discontinuity of the elongated material;
determining a walk direction of the test device;
determining at least one characteristic of the detected acoustic responses; and
determining a pattern of a feature of the elongated material associated with said discontinuity based on the determined walk direction, the at least one characteristic and at least one pattern recognition rule.
2. The method of claim 1 , wherein said at least one characteristic comprises a travel time.
3. The method of claim 1 , wherein said at least one characteristic comprises an angle of travel.
4. The method of claim 1 , wherein said at least one characteristic comprises a beam pulse speed.
5. The method of claim 1 , wherein the walk direction comprises a positive walk direction.
6. The method of claim 1 , wherein the walk direction comprises a negative walk direction.
7. The method of claim 1 , wherein the walk direction comprises a non-walk direction.
8. The method of claim 1 , wherein said step of determining the pattern comprises:
classifying features associated with the responses;
clustering the classified features;
performing basic recognition analysis using the clustered classified features to determine a basic recognition; and
performing context recognition analysis on the basic recognition to determine the pattern.
9. The method of claim 8 , wherein said step of classifying features comprises:
comparing the responses to a set of features rules; and
deriving features based on the comparisons.
10. The method of claim 9 , wherein the set of features rules are a set of rules developed by an expert.
11. The method of claim 8 , wherein said step of clustering the classified features comprises:
determining proximity relationships between the features;
comparing the proximity relationships to a set of clustering rules; and deriving clusters based on the results of the comparison.
12. The method of claim 11 , wherein the set of clustering rules are a set of rules developed by an expert.
13. The method of claim 8 , wherein said step of performing basic recognition analysis comprises:
comparing the clustered features to a set of basic recognition rules; and
deriving the basic recognition based on the results of the comparison.
14. The method of claim 13 , wherein the set of basic recognition rules are a set of rules developed by an expert.
15. The method of claim 8 , wherein said step of performing context recognition analysis comprises:
comparing the basic recognition to a set of pattern recognition rules; and
deriving the pattern based on the results of the comparison.
16. The method of claim 13 , wherein the set of pattern recognition rules are a set of rules developed by an expert.
17. The method of claim 1 , wherein the at least one pattern recognition rule comprises a multi-level expert-derived rule.
18. The method of claim 1 , wherein the elongated material is a railroad rail and the pattern is a flaw in the rail.
19. The method of claim 1 , wherein the elongated material is a railroad rail and the pattern is a feature of the rail, which is not defective.
20. The method of claim 1 , wherein said pattern comprises multiple rail features.
21. A method of performing real-time pattern recognition on a railroad rail, said method comprising the steps of:
developing a set of pattern recognition rules associated with known discontinuities within the rail;
injecting, using a test device, pulses of a plurality of ultrasonic beams into the rail;
detecting acoustic responses caused at one or more discontinuities of the rail;
determining at least one characteristic of the detected acoustic responses; and
determining a pattern of at least one feature of the rail associated with the one or more discontinuities based on the at least one characteristic and at least one pattern recognition rule.
22. The method of claim 21 , wherein said step of developing a set of pattern recognition rules comprises:
developing a set of features analysis rules;
developing a set of clustering analysis rules;
developing a set of basic recognition analysis rules; and
developing a set of context recognition analysis rules.
23. The method of claim 22 , wherein said step of determining the pattern comprises:
classifying features associated with the responses based on said feature analysis rules;
clustering the classified features based on said clustering analysis rules;
performing basic recognition analysis using the clustered classified features to determine a basic recognition based on said basic recognition analysis rules; and
performing context recognition analysis on the basic recognition to determine the pattern of the feature based on the context recognition analysis rules.
24. The method of claim 22 , wherein said rules are multi-level analysis rules.
25. The method of claim 21 further comprising:
determining a walk direction of the test device; and
using the determined walk direction to in determining the pattern.
26. The method of claim 25 , wherein the walk direction comprises a positive walk direction.
27. The method of claim 25 , wherein the walk direction comprises a negative walk direction.
28. The method of claim 25 , wherein the walk direction comprises a non-walk direction.
29. The method of claim 21 , wherein the pattern is a flaw in the rail.
30. The method of claim 21 , wherein the pattern is a feature of the rail, which is not defective.
31. The method of claim 21 , wherein the patter is an unknown flaw in the rail and said method further comprises the step of creating at least one rule associated with the unknown flaw.
32. The method of claim 21 further comprising the step of modifying the pattern recognition rules.
33. A system for performing real-time pattern recognition on an elongated material, said system comprising:
means for injecting pulses of a plurality of ultrasonic beams into the elongated material;
means for detecting acoustic responses caused at a discontinuity of the elongated material;
means for determining a walk direction of said injecting means;
means for determining at least one characteristic of the detected acoustic responses; and
means for determining a pattern of a feature of the elongated material associated with said discontinuity based on the determined walk direction, the at least one characteristic and at least one pattern recognition rule.
34. The system of claim 33 , wherein said at least one characteristic is selected from the group consisting of travel time, angle of travel, and beam pulse speed.
35. The system of claim 33 , wherein the walk direction comprises a positive walk direction.
36. The system of claim 33 , wherein the walk direction comprises a negative walk direction.
37. The system of claim 33 , wherein the walk direction comprises a non-walk direction.
38. The system of claim 33 , wherein said means for determining the pattern comprises:
means for classifying features associated with the responses;
means for clustering the classified features;
means for performing basic recognition analysis using the clustered classified features to determine a basic recognition; and
means for performing context recognition analysis on the basic recognition to determine the pattern of the feature of the elongated material associated with said discontinuity.
39. The system of claim 38 , wherein said classifying features means comprises:
means for comparing the responses to a set of features rules; and
means for deriving features based on the comparisons.
40. The system of claim 38 , wherein said clustering means comprises:
means for determining proximity relationships between the features;
means for comparing the proximity relationships to a set of clustering rules; and
means for deriving clusters based on the results of the comparison.
41. The system of claim 38 , wherein said means for performing basic recognition analysis comprises:
means for comparing the clustered features to a set of basic recognition rules; and
means for deriving the basic recognition based on the results of the comparison.
42. The system of claim 38 , wherein said means for performing context recognition analysis comprises:
means for comparing the basic recognition to a set of pattern recognition rules; and
means for deriving the pattern based on the results of the comparison.
43. The system of claim 33 , wherein the at least one pattern recognition rule comprises a multi-level expert-based rule.
44. The system of claim 33 , wherein the elongated material is a railroad rail and the pattern is a flaw in the rail.
45. The system of claim 33 , wherein the elongated material is a railroad rail and the pattern is a feature of the rail, which is not defective.
46. A system for performing real-time pattern recognition on an railroad rail, said system comprising:
a test device being adapted to ride on the rail, said test device injecting pulses of a plurality of ultrasonic beams into the rail and detecting acoustic responses caused at a discontinuity along the rail, said test device being programmed to determine a walk direction of said test device and at least one characteristic of the detected acoustic responses; and
a processor, said processor being programmed to determine a pattern of a feature of the rail associated with said discontinuity based on the determined walk direction, the at least one characteristic and at least one pattern recognition rule.