US 11,720,069 B2
Method for the computer-aided control of a technical system, more particularly a power generation installation
Daniel Hein, Munich (DE); Alexander Hentschel, Vancouver (CA); and Steffen Udluft, Eichenau (DE)
Assigned to SIEMENS AKTIENGESELLSCHAFT, Munich (DE)
Appl. No. 16/766,564
Filed by Siemens Aktiengesellschaft, Munich (DE)
PCT Filed Oct. 11, 2018, PCT No. PCT/EP2018/077663
§ 371(c)(1), (2) Date May 22, 2020,
PCT Pub. No. WO2019/101422, PCT Pub. Date May 31, 2019.
Claims priority of application No. 17203614 (EP), filed on Nov. 24, 2017.
Prior Publication US 2020/0348631 A1, Nov. 5, 2020
Int. Cl. G05B 13/04 (2006.01); G05B 13/02 (2006.01); H02J 3/38 (2006.01)
CPC G05B 13/042 (2013.01) [G05B 13/0265 (2013.01); H02J 3/38 (2013.01); H02J 2203/20 (2020.01)] 7 Claims
OG exemplary drawing
 
1. A method for computer-aided control of a power plant for power generation, to achieve a prescribed technical response from the power plant, the method comprising:
providing an operating data record for controlling the power plant, wherein the operating data record includes a respective value for a control parameter for multiple states of the power plant;
using the operating data record as a basis for using regression training to ascertain a system model for describing a manner of operation of the power plant;
determining, using a gradient-free optimization method and the system model and the operating data record, a control data record including at least one optimal value for the control parameter for at least one of the multiple states of the power plant;
taking the determined control data record as a basis for using a selection method by using a reward function to select relevant parameters of the power plant that allow more advantageous control of the power plant than other parameters of the power plant; and
using a model-based machine learning method, ascertaining a control method for the power plant by using the selected relevant parameters, wherein the control method is used to control the power plant.