SOFTWARE-CENTRIC METHODOLOGY FOR VERIFICATION AND VALIDATION OF FAULT MODELS

• US 20110145026A1
• Filed: 12/10/2009
• Published: 06/16/2011
• Est. Priority Date: 12/10/2009
• Status: Active Grant

• Current Assignee: GM Global Technology Operations LLC (General Motors Company)
• Original Assignee: GM Global Technology Operations Incorporated (Motors Liquidation Co. GUC Trust)
• Inventors: Holland, Steven W., Singh, Satnam
• Granted Patent: US 8,473,330 B2
• Time in Patent Office: Days
• US Class Current: 705/7.11
• CPC Class Codes: G06Q 10/00 Administration; Management
• CPC Class Codes: G06Q 10/063 Operations research, analysis or management
• CPC Class Codes: G06Q 10/0639 Performance analysis of employees; Performance analysis of enterprise or organisation operations

Abstract

A method for verifying and improving a vehicle fault model is disclosed. The method includes analyzing the available field failure data that includes vehicle symptoms and failures for many vehicles. The method performs an analysis using the field failure data that includes using subject matter expert knowledge to determine the most significant failure modes and the most significant symptoms. The method also includes learning simulation parameters from the field failure data and simulating faults using the learned simulation parameters. The method further includes verifying and validating the fault model based on the most significant failure modes and the most significant symptoms from the what-if analysis and the faults identifies by the simulation, and using a diagnostic reasoner to analyze the revised fault model to generate estimated faults. The method then compares the estimated faults to the simulated faults to determine true detection and false alarm rates for a benefit analysis.

Claims

1. A method for verifying, validating and improving a fault model, said method comprising:

providing an initial fault model that identifies a relationship between symptoms in a vehicle and failure modes in the vehicle;

providing field failure data that includes vehicle symptoms and vehicle failure modes of many vehicles that are in the field being operated;

performing a what-if analysis using the field failure data that includes using subject matter expert (SME) knowledge to determine the most significant failure modes and the most significant symptoms;

learning simulation parameters from the field failure data;

simulating faults using the learned simulation parameters;

revising the initial fault model using the results of the what-if analysis and the simulations;

using a diagnostic reasoner to analyze the revised fault model to generate estimated faults;

comparing the estimated faults to the simulated faults to determine true detection and false alarm rates; and

performing a benefit analysis by relating the true detection and false alarm rate to costs.

2. The method according to claim 1 wherein the field failure data includes warranty claims data, diagnostic trouble codes and parameter identifiers.

3. The method according to claim 1 wherein performing the what-if analysis includes determining a predetermined number of the most significant failure modes according to frequency of occurrence, cost and customer walk-home occurrences.

4. The method according to claim 1 wherein performing the what-if analysis includes determining a predetermined number of the most significant symptoms according to frequency of occurrence and severity.

5. The method according to claim 1 wherein learning simulation parameters from the field failure data includes identifying a bivariate failure distribution of major faults, determining the average labor cost, other labor hour cost, part cost and total cost of components repairs, determining repeat visits and multi-claim rates, determining conditional probabilities among failure modes and symptoms, determining fault appearance and disappearance probabilities for simulating intermittent faults, and learning occurrence count and severity of symptoms.

6. The method according to claim 1 wherein simulating faults includes performing a Monte Carlo simulation.

7. The method according to claim 1 wherein simulating faults using the learned parameters includes simulating permanent faults and simulating intermittent faults.

8. The method according to claim 7 wherein simulating permanent faults includes employing bivariate failure distribution to simulate faults with realistic scenarios and simulating intermittent faults includes employing fault appearance and disappearance probability distributions to simulate the intermittent faults in a realistic manner during an actual incidence and in a service bay.

9. The method according to claim 1 wherein performing the benefit analysis includes computing a decrease in misdiagnosis rate, repeat visit and multi-claim rate using diagnostic reasoning results.

10. The method according to claim 1 further comprising using a plurality of diagnostic reasoners and benchmarking the diagnostic reasoners by generating simulations and analyze them through the diagnostic reasoners.

11. A method for verifying, validating and improving a fault model, said method comprising:

providing an initial fault model that identifies a relationship between symptoms in a vehicle and failure modes in the vehicle;

providing field failure data that includes vehicle symptoms and vehicle failures of many vehicles that are in the field being operated, where the field failure data includes warranty claims data, diagnostic trouble codes and operating parameter identifies;

performing a what-if analysis using the field failure data that includes using subject matter expert (SME) knowledge to determine the most significant failure modes and the most significant systems, where the most significant failure modes are determined according to frequency of occurrence, cost and customer walk-home occurrences and determining the most significant systems according to frequency of occurrence and severity;

learning simulation parameters from the field failure data;

simulating faults using the learned simulation parameters that includes simulating permanent faults and intermittent faults;

revising the initial fault model using the results of the what-if analysis and the simulated faults;

using a diagnostic reasoner to provide a ranked order list of the estimated failure modes according to likelihood values;

comparing the estimated faults to the simulated faults to determine true detection and false alarm rates; and

performing a benefit analysis using the true detection and false alarm rate that includes computing savings due to the fault model and diagnostic reasoning.

12. The method according to claim 11 wherein learning simulation parameters from the field failure data includes identifying a bivariate failure distribution of major faults, determining the average labor cost, other labor hour cost, part cost and total cost of components repairs, determining repeat visits and multi-claim rates, determining conditional probabilities among failure modes and symptoms, determining fault appearance and disappearance probabilities for simulating intermittent faults, and learning occurrence count and severity of symptoms.

13. The method according to claim 11 wherein simulating faults includes performing a Monte Carlo simulation.

14. The method according to claim 11 wherein simulating permanent faults includes employing bivariate failure distribution to simulate faults with realistic scenarios and simulating intermittent faults includes employing faults includes employing fault appearance and disappearance probability distributions to simulate the intermittent faults in a realistic manner during an actual incidence and in a service bay.

15. The method according to claim 11 wherein performing the benefit analysis includes computing a decrease in misdiagnosis rate, repeat visit and multi-claim rate using diagnostic reasoning results.

16. A method for verifying and validating a fault model, said method comprising:

providing field failure data that includes information relating to symptoms and failures in a system;

performing a what-if analysis using the field failure data that includes using subject matter expert (SME) knowledge to determine the most significant failure modes and the most significant symptoms;

learning simulation parameters from the field failure data;

simulating faults using the learned simulation parameters; and

verifying and validating the fault model using the most significant failure modes and the most significant symptoms from the what-if analysis and the simulated faults.

17. The method according to claim 16 further comprising using a diagnostic reasoner to analyze the symptoms in the what-if scenarios and simulations to provide a ranked order list of the estimated failure modes according to likelihood values and comparing the estimated faults to the simulated faults to determine true detection and false alarm rates.

18. The method according to claim 16 wherein performing the what-if analysis includes determining a predetermined number of the most significant failure modes according to frequency of occurrence, cost and customer walk-home occurrences.

19. The method according to claim 16 wherein performing the what-if analysis includes determining a predetermined number of the most significant symptoms according to frequency of occurrence and severity.

20. The method according to claim 16 wherein simulating faults using the learned parameters includes simulating permanent faults and simulating intermittent faults.