Fault diagnostics

• US 7,925,608 B2
• Filed: 03/03/2006
• Issued: 04/12/2011
• Est. Priority Date: 03/18/2005
• Status: Active Grant

• Current Assignee: British Telecommunications PLC (BT Group PLC)
• Original Assignee: British Telecommunications PLC (BT Group PLC)
• Inventors: Rohlfing, David M
• Primary Examiner(s): HOLMES, MICHAEL B
• Application Number:
  US11/885,752
• Publication Number:
  US 20080168018A1
• Time in Patent Office: 1,866 Days
• Field of Search: 706/47
• US Class Current: 706/47
• CPC Class Codes: H04L 41/16 using machine learning or artificial intelligence

Abstract

A method of fault diagnostics in a case based reasoning system, wherein said case based reasoning system comprises a plurality of cases, each case having an associated solution and a case weight, said method comprising the steps of: receiving data associated with a fault; determining a case match indicative of the degree of matching between the received data and each of the plurality of cases; providing a solution associated with one of the plurality of cases in dependence on its case match and case weight; receiving an actual solution to the fault; increasing the case weight of the case associated with the provided solution if the actual solution is equal to the provided solution; and adding a new case to the plurality of cases if the actual solution is not equal to the provided solution, wherein the actual solution is associated with the new case.

Claims

1. A method of fault diagnostics in a case based reasoning system, wherein said case based reasoning system comprises a plurality of cases, each case having an associated solution and a case weight parameter indicative of the reliability of the associated solution, said method comprising the steps of:

i) receiving data associated with a fault in a telecommunications network;

ii) determining a case match score indicative of the degree of matching between the received data and each of the plurality of cases;

iii) calculating a weighted case match score based on the case match score and the associated case weight parameter for each of the plurality of cases, and then providing a solution associated with one of the plurality of cases in dependence on the weighted case match scores and case weight;

iv) receiving an actual solution to the fault, and

a) increasing the case weight parameter of the case associated with the provided solution if the actual solution is equal to the provided solution;

or

b) adding a new case to the plurality of cases if the actual solution is not equal to the provided solution, wherein the actual solution is associated with the new case; and

v) decaying over time the case weight parameters associated with each of the plurality of cases.

2. A method of fault diagnostics according to claim 1, wherein the decay over time is periodic.

3. A method of fault diagnostics according to claim 1, further comprising normalising the results from step ii).

4. A method of fault diagnostics according to claim 3, wherein the normalising comprises dividing the weighted case match score for a case by the sum of the all case weight parameters corresponding to the solution associated with said case.

5. A method of fault diagnostics according to claim 1, wherein steps i) to vi) are repeated.

6. A method of fault diagnostics according to claim 1, wherein the higher the weighted case match score of a case, the better the match between the case and the fault.

7. A case based reasoning system for fault diagnostics comprising a processor, an input module adapted for receiving data associated with a fault, storage means storing a plurality of cases, each case having an associated solution and a case weight parameter indicative of the reliability of the associated solution, wherein the processor is configured so as to perform:

i) determining a case match score indicative of the degree of matching between the received data and each of the plurality of cases;

ii) calculating a weighted case match score based on the case match score and the associated case weight parameter for each of the plurality of cases, and providing a solution associated with one of the plurality of cases in dependence on its weighted case match score;

iii) receiving an actual solution to the fault; and

a) increasing the case weight parameter of the case associated with the provided solution if the actual solution is equal to the provided solution;

or

b) adding a new case to the plurality of cases if the actual solution is not equal to the provided solution, wherein the actual solution is associated with the new case; and

iv) decaying over time the case weight parameters associated with each of the plurality of cases.

8. A system according to claim 7, wherein the decay over time is periodic.

9. The system according to claim 7, wherein the processor is further configured to normalize the results from step ii).

10. The system according to claim 9, wherein the normalizing comprises dividing the weighted case match score for a case by the sum of the all case weight parameters corresponding to the solution associated with said case.

11. The system according to claim 7, wherein the processor is further configured to repeat steps i) to vi).

12. The system according to claim 7, wherein the higher the weighted case match score of a case, the better the match between the case and the fault.