Method of modeling vehicle parameter cycles

• US 20060155541A1
• Filed: 01/07/2005
• Published: 07/13/2006
• Est. Priority Date: 01/07/2005
• 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: Tate, Edward D. Jr.
• Granted Patent: US 7,640,148 B2
• Time in Patent Office: Days
• US Class Current: 704/256.300
• CPC Class Codes: B60W 20/11 using model predictive control [MPC] strategies, i.e. control methods based on models predicting performance utilising navigation and traffic information in the control strategy B60W20/12
• CPC Class Codes: B60W 2050/0025 Transfer function weighting factor
• CPC Class Codes: B60W 2050/0028 Mathematical models, e.g. for simulation
• CPC Class Codes: B60W 2050/0057 Frequency analysis, spectral techniques or transforms
• CPC Class Codes: B60W 40/12 related to parameters of the vehicle itself , e.g. tyre models

Abstract

A method of modeling a cycle of a vehicle parameter includes generating a plurality of training sets and partially classifying elements of each training set based on a present value of the vehicle parameter and a subsequent change in the vehicle parameter. The method further includes reducing the plurality of training sets to a corresponding plurality of parameter sets based on the elements and defining a plurality of statistical groups of the parameter sets based on a value and a history of that value.

Claims

1. A method of modeling a cycle of a vehicle parameter, comprising:

generating a plurality of training sets;

partially classifying elements of each training set based on a present value of said vehicle parameter and a subsequent change in said vehicle parameter;

reducing said plurality of training sets to a corresponding plurality of parameter sets based on said elements; and

defining a plurality of statistical groups of said parameter sets based on a value and a history of that value.

2. The method of claim 1 wherein said plurality of training sets each include a plurality of parameter vectors.

3. The method of claim 1 further comprising weighting each of said training sets whereby more recent elements of a particular training set have a greater influence on said statistical groups than earlier elements of said particular training set.

4. The method of claim 1 further comprising approximating each of said training sets based on respective lower order representations of each training set.

5. The method of claim 1 wherein said step of defining a plurality of statistical groups comprises performing cluster analysis on said training sets to generate said statistical groups.

6. The method of claim 5 wherein said cluster analysis is performed based on said value and a plurality of historical values.

7. A method of determining a future state of a stochastic system, comprising:

generating a plurality of training sets;

partially classifying elements of each training set based on a history of a state of said stochastic system;

reducing said plurality of training sets to a corresponding plurality of parameter sets based on said elements;

defining a plurality of statistical groups of said state; and

assigning a present state of said stochastic system to one of said statistical groups; and

determining said future state based on a state probability distribution generated based on said present state and said statistical group.

8. The method of claim 7 wherein said plurality of training sets each include a plurality of parameter vectors.

9. The method of claim 7 further comprising weighting each of said training sets whereby more recent elements of a particular training set have a greater influence on said statistical groups than earlier elements of said particular training set.

10. The method of claim 7 further comprising approximating each of said training sets based on respective lower order representations of each training set.

11. The method of claim 7 wherein said step of defining a plurality of statistical groups comprises performing cluster analysis on said training sets to generate said statistical groups.

12. The method of claim 11 wherein said cluster analysis is performed based on a value of said state and a plurality of historical values of said state.

13. A method of regulating operation of a vehicle, comprising:

determining a current vehicle parameter value;

assigning said current vehicle parameter value to one of a plurality of statistical groups defined by a conditional probability model (CPM) of said vehicle;

calculating a statistical distribution of a future vehicle parameter value; and

regulating operation of said vehicle based on a most likely value of said future vehicle parameter.

14. The method of claim 13 further comprising updating said CPM based on said current vehicle parameter.

15. The method of claim 13 wherein said step of assigning said current vehicle parameter value, comprises:

determining a center point value of each of said statistical groups; and

assigning said current vehicle parameter value to said one of said plurality of said statistical groups having said center point value closest to said current vehicle parameter value.

16. The method of claim 13 further comprising generating said CPM of said vehicle.

17. The method of claim 16 wherein said step of generating said CPM comprises:

generating a plurality of training sets;

partially classifying elements of each training set based on said vehicle parameter value and a subsequent change in said vehicle parameter value;

reducing said plurality of training sets to a corresponding plurality of parameter sets based on said elements; and

defining said plurality of statistical groups of said parameter sets based on said vehicle parameter value and a history of that vehicle parameter value.

18. The method of claim 17 wherein said plurality of training sets each include a plurality of parameter vectors.

19. The method of claim 17 further comprising weighting each of said training sets whereby more recent elements of a particular training set have a greater influence on said statistical groups than earlier elements of said particular training set.

20. The method of claim 17 further comprising approximating each of said training sets based on respective lower order representations of each training set.

21. The method of claim 17 wherein said step of defining said plurality of statistical groups comprises performing cluster analysis on said training sets to generate said statistical groups.

22. The method of claim 21 wherein said cluster analysis is performed based on said vehicle parameter value and a plurality of historical values.