DIAGNOSIS IN AUTOMOTIVE APPLICATIONS

• US 20070283188A1
• Filed: 04/13/2007
• Published: 12/06/2007
• Est. Priority Date: 04/13/2006
• Status: Active Grant

• Current Assignee: Dspace GmbH
• Original Assignee: Dspace Digital Signal Processing And Control Engineering Gmbh
• Inventors: Balzer, Heinrich, Niggemann, Oliver, Stein, Benno Maria
• Granted Patent: US 7,991,583 B2
• Time in Patent Office: Days
• US Class Current: 714/26
• CPC Class Codes: B60W 2050/0014 Adaptive controllers
• CPC Class Codes: B60W 2050/0031 Mathematical model of the vehicle
• CPC Class Codes: B60W 2050/0037 Mathematical models of vehicle sub-units
• CPC Class Codes: B60W 2050/009 Priority selection
• CPC Class Codes: B60W 50/0205 Diagnosing or detecting failures; Failure detection models
• CPC Class Codes: B60W 50/045 Monitoring control system parameters
• CPC Class Codes: F02D 11/107 Safety-related aspects
• CPC Class Codes: F02D 2041/1433 using a model or simulation of the system
• CPC Class Codes: F02D 2041/1437 Simulation
• CPC Class Codes: F02D 2200/602 Pedal position
• CPC Class Codes: F02D 41/1405 Neural network control
• CPC Class Codes: F02D 41/22 Safety or indicating devices for abnormal conditions in air/fuel ratio feedback systems F02D41/1495, in electric control linkage F02D11/107, in purge control systems F02M25/0809

Abstract

A system and method for constructing a diagnosis function for mapping a symptom of a fault in an electromechanical system onto the fault by simulating the electromechanical system in at least a fault mode and a node fault mode, and learning a classifier function from the collected simulation results for mapping the system of the fault onto the fault.

Claims

1. A method for constructing a diagnosis function for mapping a symptom of a fault in an electromechanical system onto the fault, the method comprising the steps of:

simulating the electromechanical system in at least a fault-mode and a non-fault-mode; and

learning a classifier function from collected simulation results for mapping the symptom of the fault onto the fault.

2. The method according to claim 1, wherein the simulating step comprises simulating the electromechanical system for given vectors of input functions, and further comprising the step of determining symptom vectors by compiling the results of fault- and non-fault-simulations and comparing related results.

3. The method according to claim 1, wherein the learning step includes at least one of filtering, machine learning, data mining, and pattern recognition.

4. The method according to claim 1, further comprising the step of:

prior to the learning step, generalizing the symptom.

5. The method according to claim 1, wherein the electromechanical system is a mechatronic system, and the method further comprises the steps of:

detecting the symptom of the fault in the mechatronic system; and

mapping the detected symptom onto the fault using the learned classifier function.

6. The method according to claim 5, wherein the detecting step comprises detecting the system by comparing an expected behavior with an observed behavior of the mechatronic system.

9. The method according to claim 1 wherein the fault causes the symptom.

10. The method according to claim 1, wherein the learning step is performed using linear regression.

11. The method according to claim 1, wherein the learning step is performed using decision trees.

12. The method according to claim 1, wherein the electromechanical system includes a vehicle.

13. The method according to claim 2, further comprising the step of:

generalizing the compiled symptom vectors prior to the learning step.

14. The method according to claim 13, wherein the generalizing step comprises abstracting the compiled symptom vectors into intervals.

15. The method according to claim 6, wherein the detecting step comprises using a sensor reading.

7. An apparatus for detecting a fault in a mechatronic system, the apparatus comprising at least one electronic control unit operable to run a learned classifier function, to detect a symptom of a fault, and to map the system onto the fault using the classifier function.

8. The apparatus according to claim 7, further comprising:

at least one sensor operable to provide a sensor reading,

wherein the electronic control unit is operable to detect the symptom by comparing an expected behavior with an observed behavior of the system using the sensor reading.

16. The apparatus according to claim 7, wherein the mechatronic system includes a vehicle.

17. A method for constructing a diagnosis function for mapping a symptom of a fault in an electromechanical system onto the fault causing the symptom, the method comprising the steps of:

simulating the electromechanical system in a fault-mode;

generating a first set of simulation results;

storing the first set of simulation results in a database;

simulating the electromechanical system in a non-fault mode;

generating a second set of simulation results;

storing the second set of simulation results in the database;

comparing the first and second sets of simulation results to detect patterns in the results;

developing a classification function which determines a cause of the fault;

applying the classification function to map the determined fault onto the fault causing the symptom.

18. The method according to claim 17, further comprising the steps of:

simulating the electromechanical system for a set of vectors of an input function; and

determining a set of symptom vectors by compiling the results of the fault-mode and non-fault mode simulations and comparing related results.

19. The method according to claim 17, further comprising the step of:

developing the classification function by learning, using at least one of filtering, machine learning, data mining, and pattern recognition.

20. The method according to claim 17, wherein the electromechanical system includes a vehicle.

21. The method according to claim 17, further comprising the step of:

detecting symptoms by comparing an expected behavior with an observed behavior of the electromechanical system.