Fault prediction framework using temporal data mining

• US 8,676,432 B2
• Filed: 01/13/2010
• Issued: 03/18/2014
• Est. Priority Date: 01/13/2010
• Status: Active Grant

• Current Assignee: GM Global Technology Operations LLC (General Motors Company)
• Original Assignee: GM Global Technology Operations LLC (General Motors Company)
• Inventors: Patnaik, Debprakash, Bandyopadhyay, Pulak, Holland, Steven W., Unnikrishnan, Kootaala P., Montgomery, George Paul Jr.
• Primary Examiner(s): Black, Thomas
• Assistant Examiner(s): Lewandroski, Sara
• Application Number:
  US12/686,963
• Publication Number:
  US 20110172874A1
• Time in Patent Office: 1,525 Days
• Field of Search: 701/29.1, 701/29.9, 701/31.6, 701/31.8, 701/31.9, 701/32.9, 701 335- 337
• US Class Current: 701/29.1
• CPC Class Codes: G06F 11/008 Reliability or availability analysis
• CPC Class Codes: G06F 16/2474 Sequence data queries, e.g. querying versioned data

Abstract

A vehicle fault diagnosis and prognosis system includes a computing platform configured to receive a classifier from a remote server, the computing platform tangibly embodying computer-executable instructions for evaluating data sequences received from a vehicle control network and applying the classifier to the data sequences, wherein the classifier is configured to determine if the data sequences define a pattern that is associated with a particular fault.

Claims

1. A vehicle fault diagnosis and prognosis system, comprising:

a computing platform on the vehicle configured to receive a fault classifier from a remote server, the computing platform tangibly embodying computer-executable instructions for;

evaluating local data sequences received from a vehicle control network; and

applying the classifier to the local data sequences, wherein the classifier is configured to determine if the local data sequences define a pattern associated with a particular fault where the classifier is trained by comparing data sequences from a first population of vehicles to repair data from a second population of vehicles.

2. The system of claim 1, wherein the local data sequences include one or more the following:

sequenced diagnostic trouble codes, time stamped diagnostic trouble codes and vehicle parameters.

3. The system of claim 1, wherein the classifier is trained to associate a plurality of data sequence clusters to a particular failure mode.

4. The system of claim 1, wherein the data sequences in the classifier are derived from the first population of vehicles using temporal data mining.

5. A method for fault diagnosis and prognosis, comprising:

extracting data sequences from a first population of vehicles;

applying a temporal data mining application to the data sequences extracted from the first population of vehicles to detect patterns in the data sequences;

retrieving repair data related to a second population of vehicles;

comparing the data sequences to the repair data to identify data sequences that are related to a particular failure mode;

applying a clustering algorithm to create clusters of data sequences related to the particular failure mode using the identified data sequences that are related to the particular failure mode;

training a fault prediction classifier to learn the particular failure modes associated with each cluster so the fault prediction classifier can identify the data sequences that lead to particular failure modes, said fault prediction classifier using a neural network, a support vector machine and a decision tree;

sending the fault classifier to a computing platform on a vehicle; and

receiving the fault classifier in the computing platform and the computing platform executing instructions for;

evaluating data sequences received from a vehicle control network; and

applying the classifier to the data sequences, wherein the classifier is configured to determine if the data sequences define a pattern associated with a particular fault.

6. The method of claim 5, wherein applying the clustering algorithm includes assigning a probability to each data sequence and particular failure mode combination.

7. The method of claim 5, wherein applying the clustering algorithm includes clustering the data sequences according to a particular failure mode.

8. The method of claim 5, wherein applying the temporal data mining application includes identifying frequent data sequences.

9. The method of claim 5, wherein the first population of vehicles is similar to the second population of vehicles.

10. The method of claim 5, wherein applying the clustering algorithm to create clusters includes assigning a confidence measure to each of the identified data sequences related to a particular failure mode.

11. A computer-readable medium tangibly embodying computer-executable instructions for:

extracting data sequences from a first population of vehicles;

applying a temporal data mining application to the data sequences extracted from the first population of vehicles to detect patterns in the data sequences;

retrieving repair data related to a second population of vehicles;

comparing the data sequences to the repair data to identify data sequences that are related to a failure mode;

applying a clustering algorithm to create clusters of data sequences related to the failure mode using the identified data sequences that are related to the failure mode; and

training a fault prediction classifier to learn the failure modes associated with each cluster so the fault prediction classifier can identify the data sequences that lead to failure modes, said fault prediction classifier using a neural network.

12. The computer-readable medium of claim 11, wherein the fault prediction classifier uses a support vector machine and a decision tree.

13. The computer-readable medium of claim 11, wherein the fault prediction classifier uses a decision tree.

14. The computer-readable medium of claim 11, wherein the fault prediction classifier uses a support vector machine.

15. The computer-readable medium of claim 11, wherein applying the temporal data mining application includes identifying frequent data sequences.

16. The computer-readable medium of claim 11, wherein applying the clustering algorithm to create clusters includes assigning a confidence measure to each of the identified data sequences related to a failure mode.

17. The computer-readable medium of claim 11, wherein applying the clustering algorithm includes clustering the data sequences according to a failure mode.

18. The computer-readable medium of claim 11, wherein applying the clustering algorithm includes assigning a probability to each data sequences and failure mode combination.

19. The computer-readable medium of claim 18, further including downloading the classifier to an off-board computing platform.

20. The computer-readable medium of claim 18, further including downloading the classifier to an on-board flexible computing platform.

21. The computer-readable medium of claim 11, wherein the clustering algorithm is a distance measuring algorithm.