Method and device for creating a function model for a control unit of an engine system

Patent metadata

Field	Value
Publication number	US-10339463-B2
Application number	US-201414247121-A
Country	US
Kind code	B2
Filing date	Apr 7, 2014
Priority date	Apr 10, 2013
Publication date	Jul 2, 2019
Grant date	Jul 2, 2019

Abstract

A computerized method for creating a function model based on a non-parametric, data-based model, e.g., a Gaussian process model, includes: providing training data including measuring points having one or multiple input variables, the measuring points each being assigned an output value of an output variable; providing a basic function; modifying the training data with the aid of difference formation between the function values of the basic function and the output values at the measuring points of the training data; creating the data-based model based on the modified training data; and providing the function model as a function of the data-based model and the basic function.

First claim

What is claimed is:

1. A method, performed via a processor, for generating a function model, for emulating route and system functions in a control unit for a combustion engine system, based on a non-parametric, data-based model, the method comprising:
providing training data including measuring points having at least one input variable, the measuring points each being assigned an output value of an output variable;
providing a basic function;
modifying the training data based on a difference between the function values of the basic function and the output values at the measuring points of the training data;
generating the data-based model based on the modified training data;
generating the function model as a function of the data-based model and the basic function; and
providing the function model, for emulating the route and system functions in the control unit for the combustion engine system; and
controlling the combustion engine system based on the function model;
wherein the training data improves the function model used by the processor,
wherein the basic function is one of (i) predefined as a D-dimensional hyperplane, or (ii) defined by predefining target values at target value points, the basic function being defined by a number D+1 target values for a number of D input variables, and
wherein the function model provides improved control of the combustion engine system, since the function model converges in an extrapolation range against a predefined basic function.

2. The method as recited in claim 1 , wherein a validity range is predefined, the function values are also determined with the function model within the validity range and with one of the basic function or another predefined, constant function outside of the validity range.

3. The method as recited in claim 2 , wherein the validity range is ascertained as a bounding box from the training data, the bounding box defining an input variable space with axis-parallel edges, and all measuring points of the training data being located within the bounding box.

4. The method as recited in claim 3 , wherein the basic function is predefined as (i) one of a linear or constant function and (ii) one of a parametric model, a non-parametric model, or a physical model.

5. The method as recited in claim 3 , wherein the basic function corresponds to a mean value function which results from a multi-linear interpolation between the target variables.

6. A method, performed via a processor, for generating a function model, for emulating route and system functions in a control unit for a combustion engine system, based on a non-parametric, data-based model, the method comprising:
providing training data including measuring points having at least one input variable, the measuring points each being assigned an output value of an output variable;
providing a basic function;
modifying the training data based on a difference between the function values of the basic function and the output values at the measuring points of the training data;
generating the data-based model based on the modified training data;
generating the function model as a function of the data-based model and the basic function;
providing the function model, for emulating the route and system functions in the control unit for the combustion engine system; and
controlling the combustion engine system based on the function model;
wherein the training data improves the function model used by the processor,
wherein a validity range is predefined, the function values being determined with the aid of the function model within the validity range,
wherein the another predefined function corresponds to a constant function f(x)=f(min(x max , max(x min , x))) for a measuring point x, the operators “max” and “min” being used dimensionally and the vectors x max and x min including the upper and lower margins of the input variable space for each input variable, and
wherein the function model provides improved control of the combustion engine system, since the function model converges in an extrapolation range against a predefined basic function.

7. The method as recited in claim 6 , wherein a validity range is predefined, the function values are also determined with-the function model within the validity range and with one of the basic function or another predefined, constant function outside of the validity range.

8. The method as recited in claim 7 , wherein the validity range is ascertained as a bounding box from the training data, the bounding box defining an input variable space with axis-parallel edges, and all measuring points of the training data being located within the bounding box.

9. The method as recited in claim 8 , wherein the basic function is predefined as (i) one of a linear or constant function and (ii) one of a parametric model, a non-parametric model, or a physical model.

10. The method as recited in claim 8 , wherein the basic function corresponds to a mean value function which results from a multi-linear interpolation between the target variables.

11. A computing device for generating a function model, for emulating route and system functions in a control unit for a combustion engine system, based on a non-parametric, data-based model, comprising:
a processor configured to perform the following:
provide training data including measuring points having at least one input variable, the measuring points each being assigned an output value of an output variable;
provide a basic function;
modify the training data based on a difference between the function values of the basic function and the output values at the measuring points of the training data;
generate the data-based model based on the modified training data;
generate the function model as a function of the data-based model and the basic function; and
provide the function model, for emulating the route and system functions in the control unit for the combustion engine system; and
control the combustion engine system based on the function model;

wherein the training data improves the function model used by the computing device,
wherein the basic function is one of (i) predefined as a D-dimensional hyperplane, or (ii) defined by predefining target values at target value points, the basic function being defined by a number D+1 target values for a number of D input variables, and
wherein the function model provides improved control of the combustion engine system, since the function model converges in an extrapolation range against a predefined basic function.

12. The computing device as recited in claim 11 , wherein a validity range is predefined, the function values are also determined with-the function model within the validity range and with one of the basic function or another predefined, constant function outside of the validity range.

13. The computing device as recited in claim 12 , wherein the validity range is ascertained as a bounding box from the training data, the bounding box defining an input variable space with axis-parallel edges, and all measuring points of the training data being located within the bounding box.

14. The computing device as recited in claim 13 , wherein the basic function is predefined as (i) one of a linear or constant function and (ii) one of a parametric model, a non-parametric model, or a physical model.

15. The computing device as recited in claim 13 , wherein the basic function corresponds to a mean value function which results from a multi-linear interpolation between the target variables.

16. A non-transitory, computer-readable data storage medium stori

Assignees

• Bosch Gmbh Robert

Inventors

• Markert Heiner
• Diener Rene
• Kloppenburg Ernst
• Streichert Felix
• Hanselmann Michael

Classifications

• G06N20/00 Primary

  Machine learning · CPC title