Method and system for root cause analysis and quality monitoring of system-level faults

• US 8,732,112 B2
• Filed: 12/19/2011
• Issued: 05/20/2014
• Est. Priority Date: 12/19/2011
• Status: Active Grant

• Current Assignee: GM Global Technology Operations LLC (General Motors Company)
• Original Assignee: GM Global Technology Operations LLC (General Motors Company)
• Inventors: Singh, Satnam, Bowman, John V
• Primary Examiner(s): Holmes, Michael B
• Application Number:
  US13/329,816
• Publication Number:
  US 20130159240A1
• Time in Patent Office: 883 Days
• Field of Search: 706/51
• US Class Current: 706/51
• CPC Class Codes: G05B 23/0275 Fault isolation and identification, e.g. classify fault; estimate cause or root of failure
• CPC Class Codes: G07C 5/0816 Indicating performance data, e.g. occurrence of a malfunction

Abstract

System fault codes, component fault codes, and failure modes may be received at a system. A first level model comprising first level causal relationships and first level causal weights between the system fault codes and the component fault codes may be generated. A second level model comprising second level causal relationships and second level causal weights between the component fault codes and the failure modes may be generated. The first level model and the second level model may be combined to generate a model such as graphical model. One or more causal probabilities may be determined using the graphical model.

Claims

1. A method comprising:

receiving system fault codes, component fault codes, and failure modes;

generating a first level model comprising first level causal relationships and first level causal weights between the system fault codes and the component fault codes;

generating a second level model comprising second level causal relationships and second level causal weights between the component fault codes and the failure modes;

combining the first level model and the second level model to generate a graphical model; and

determining, using the graphical model, one or more probability inferences.

2. The method of claim 1, wherein:

the first level causal weights comprise conditional probabilities of the system fault codes given the component fault codes; and

the second level causal weights comprise conditional probabilities of the component fault codes given the failure modes.

3. The method of claim 1, comprising:

receiving one or more vehicle specific system fault codes and one or more vehicle specific component fault codes;

determining one or more high probability failure modes based on the probability inferences, wherein determining, using the graphical model, one or more probability inferences comprises determining conditional probabilities based on the vehicle specific system fault codes, the vehicle specific component fault codes and the graphical model; and

outputting high probability failure modes to a user.

4. The method of claim 1, comprising:

receiving a vehicle malfunction record comprising one or more fleet vehicle system fault codes, one or more fleet vehicle component fault codes, and one or more fleet vehicle failure modes;

categorizing the vehicle malfunction record based on the probability inferences, wherein determining, using the graphical model, one or more probability inferences comprises determining marginal probabilities based on the fleet vehicle system fault codes, the fleet vehicle component fault codes, and the fleet vehicle failure modes; and

outputting the categorized vehicle malfunction record to a user.

5. The method of claim 4, wherein outputting the categorized vehicle malfunction record to a user comprises outputting, if the categorized vehicle malfunction record comprises a low or medium probability, the one or more vehicle trouble records to a quality manager.

6. The method of claim 4, wherein outputting the one or more categorized vehicle trouble records to a user comprises outputting, if the categorized vehicle malfunction record comprises a probability below a threshold, the one or more vehicle trouble records to a diagnostic engineer.

7. The method of claim 1, comprising:

receiving one or more vehicle malfunction records comprising one or more fleet vehicle system fault codes, one or more fleet vehicle component fault codes, and one or more fleet vehicle failure modes; and

updating the graphical model to comprise one or more vehicle malfunction records not previously in graphical model.

8. The method of claim 1, wherein the graphical model comprises a probabilistic hierarchical model.

9. A system comprising:

a memory; and

a processor to;

receive system fault codes, component fault codes, and failure modes;

generate a first level model comprising first level causal relationships and first level causal weights between the system fault codes and the component fault codes;

generate a second level model comprising second level causal relationships and second level causal weights between the component fault codes and the failure modes;

combine the first level model and the second level model to generate a graphical model; and

determine, using the graphical model, one or more causal probabilities.

10. The system of claim 9, wherein:

the first level causal weights comprise conditional probabilities of the system fault codes given the component fault codes; and

the second level causal weights comprise conditional probabilities of the component fault codes given the failure modes.

11. The system of claim 9, wherein the processor is to:

receive one or more vehicle specific system fault codes and one or more vehicle specific component fault codes;

determine one or more high probability failure modes based on the causal probabilities, wherein to determine, using the graphical model, one or more causal probabilities the processor is to determine conditional probabilities based on the vehicle specific system fault codes, the vehicle specific component fault codes and the graphical model; and

output high probability failure modes to a user.

12. The system of claim 9, wherein the processor is to:

receive a vehicle malfunction record comprising one or more fleet vehicle system fault codes, one or more fleet vehicle component fault codes, and one or more fleet vehicle failure modes;

categorize the vehicle malfunction record based on the causal probabilities, wherein to determine, using the graphical model, one or more causal probabilities the processor is to determine marginal probabilities based on the fleet vehicle system fault codes, the fleet vehicle component fault codes, and the fleet vehicle failure modes; and

output the categorized vehicle malfunction record to a user.

13. The system of claim 12, wherein to output the categorized vehicle malfunction record to a user the processor is to output, if the categorized vehicle malfunction record comprises a low or medium probability, the one or more vehicle trouble records to a quality manager.

14. The method of claim 12, wherein to output the one or more categorized vehicle trouble records to a user the processor is to output, if the categorized vehicle malfunction record comprises a probability below a threshold, the one or more vehicle trouble records to a diagnostic engineer.

15. The system of claim 9, wherein the processor is to:

receive one or more vehicle malfunction records comprising one or more fleet vehicle system fault codes, one or more fleet vehicle component fault codes, and one or more fleet vehicle failure modes; and

update the graphical model to comprise one or more vehicle malfunction records not previously in graphical model.

16. The system of claim 9, wherein the graphical model comprises an acyclic graphical model.

17. A method comprising:

receiving system diagnostic trouble codes (DTCs), component DTCs, and root causes;

determining a first level model comprising first level causal relationships and first level causal weights between system DTCs and component DTCs;

determining a second level model comprising second level causal relationships and second level causal weights between component DTCs and root causes; and

generating an acyclic graphical model based on the first level model and the second level model.

18. The method of claim 17, comprising:

receiving one or more individual vehicle system DTCs and one or more individual vehicle component DTCs;

calculating a causal probability using the acyclic graphical model in conjunction with the individual vehicle system DTCs and one or more individual vehicle component DTCs;

determining one or more highly likely root causes based on the causal probability; and

displaying the one or more highly likely root causes to a user.

19. The method of claim 17, comprising:

receiving a vehicle trouble record comprising one or more fleet vehicle system DTCs, one or more fleet vehicle component DTCs, and one or more fleet vehicle failure modes;

calculating a marginal probability using the acyclic graphical model in conjunction with the one or more fleet vehicle system DTCs, the one or more fleet vehicle component DTCs, and the one or more fleet vehicle failure modes;

categorizing the vehicle trouble record based on the marginal probability; and

outputting the categorized vehicle trouble record to a user.

20. The method of claim 17, wherein:

the first level causal weights comprise conditional probabilities of system DTCs given component DTCs; and

the second level causal weights comprise conditional probabilities of component DTCs given root causes.