Method for calibrating a technical system

Patent metadata

Field	Value
Publication number	US-11899414-B2
Application number	US-201917311966-A
Country	US
Kind code	B2
Filing date	Dec 9, 2019
Priority date	Dec 10, 2018
Publication date	Feb 13, 2024
Grant date	Feb 13, 2024

Abstract

Various aspects of the present disclosure are directed to methods for calibrating a technical system with respect to stochastic influences during real operation of the technical system. In one example embodiment of the present disclosure, the method includes the steps of: determining the values of a number of control variables, carrying out the calibration on the basis of a load cycle which results in a sequence of a number of operating points, executing the load cycle multiple times under the influence of at least one random influencing variable, with each realization of the load cycle resulting in a random sequence of the number i of operating points, defining a risk measure of the probability distribution, with which the probability distribution is mapped to a scalar variable, and optimizing the risk measure by varying the number of control variables in order to obtain optimal control variables for calibration.

First claim

The invention claimed is:

1. Method for calibrating a technical system including the steps of:
determining the values of a number of control variables of the technical system, with which the technical system is controlled or tuned, at specific operating points of the technical system,
carrying out the calibration on the basis of a load cycle for the technical system which results in a sequence of a number of operating points of the technical system,
executing the load cycle multiple times under the influence of at least one random influencing variable, with each realization of the load cycle resulting in a random sequence of the number i of operating points,
defining a cost function which contains a target function y with a model f for an output variable of the technical system, the model f being dependent on the number of control variables of the technical system and on the number i of random operating points of the technical system so that the value of the cost function for each realization of the load cycle is itself a random variable that has a probability distribution,
defining a risk measure p of the probability distribution, with which the probability distribution is mapped to a scalar variable, and
optimizing the risk measure ρ by varying the number of control variables in order to obtain the optimal control variables for calibration.

2. The method according to claim 1 , further including the step of simulating the load cycle multiple times under the influence of the at least one random influencing variable in order to obtain the random sequences of the number i of operating points.

3. The method according to claim 1 , further including the step of operating the technical system multiple times for the load cycle under the influence of the at least one random influencing variable in order to obtain the random sequences of the number i of operating points.

4. The method according to claim 1 , further including the step of weighting a model value of the model fin the target function y with a predetermined weighting wfi and/or with predetermined costs cf.

5. The method according to claim 1 , wherein the step of defining the cost function further includes taking into account at least one restriction Bk with a model gk for an output variable of the technical system in the cost function, and wherein the model gk is dependent on the number of control variables of the technical system and on the number i of random operating points of the technical system.

6. The method according to claim 5 , further including the step of weighting a model value of the model gk of the at least one restriction Bk with a predetermined weight wgk and/or with predetermined costs cgk.

7. The method according to claim 5 , further including the step of evaluating a violation of a restriction Bk with a predetermined penalty function φ.

8. The method according to claim 1 , wherein the step of optimizing further includes taking into account a stochastic constraint for at least one restriction Bk with a model gk for an output variable of the technical system, wherein the model gk is dependent on the number of control variables of the technical system and on the number i of random operating points of the technical system.

9. The method according to claim 8 , further including the steps of
determining the probability W that the stochastic constraint exceeds a predetermined limit value Gk, and
comparing the probability W to a predetermined probability bound WS.

10. The method according to claim 1 , wherein the step of optimizing further includes varying iteratively the number of control variables and determining again the probability distribution of the cost function until the risk measure p of the probability distribution is optimized.

11. The method according to claim 1 , wherein the step of optimizing further includes taking into account a non-stochastic boundary condition and/or constraint for the number of control variables.

Assignees

• Avl List Gmbh

Inventors

• Didcock Nico
• Wasserburger Alexander
• Hametner Christoph

Classifications

• G05B13/041 Primary

  in which a variable is automatically adjusted to optimise the performance · CPC title

• G05B1/03 Primary

  for comparing digital signals · CPC title

• G06F17/18

  for evaluating statistical data {, e.g. average values, frequency distributions, probability functions, regression analysis (forecasting specially adapted for a specific administrative, business or logistic context G06Q10/04)} · CPC title

• F02D41/2432 Primary

  Methods of calibration · CPC title

• F02D41/1405

  Neural network control · CPC title