	US 12,234,740 B2
	Training of machine learning models for data-driven decision-making
Shankar Deepak Srinivasan, Berlin (DE); Klaus Paul, Berlin (DE); Shri Nishanth Rajendran, Berlin (DE); and Astrid Walle, Berlin (DE)
Assigned to ROLLS-ROYCE DEUTSCHLAND LTD & CO KG, Blankenfelde-Mahlow (DE)
Appl. No. 17/761,484
Filed by ROLLS-ROYCE DEUTSCHLAND LTD & CO KG, Blankenfelde-Mahlow (DE)
PCT Filed Sep. 23, 2020, PCT No. PCT/EP2020/076504 § 371(c)(1), (2) Date Mar. 17, 2022, PCT Pub. No. WO2021/058526, PCT Pub. Date Apr. 1, 2021.
Claims priority of application No. 10 2019 214 653.8 (DE), filed on Sep. 25, 2019.
Prior Publication US 2022/0364478 A1, Nov. 17, 2022
Int. Cl. F01D 21/00 (2006.01); G06N 20/00 (2019.01)

CPC F01D 21/003 (2013.01) [F05D 2220/32 (2013.01); F05D 2260/80 (2013.01); G06N 20/00 (2019.01)]

17 Claims

1. A method for training machine learning models, comprising:

capturing, by one or more computers, of data obtained by one or more measuring devices, each being a sensor for measuring a physical quantity as time series data;

receiving, by the one or more computers, multiple classification data units relating to the data;

receiving, by the one or more computers and for each classification data unit, a selected portion of the data; and

training, by the one or more computers, multiple machine learning models, each on a basis of at least one of the classification data units and the applicable selected portion of the data, wherein the multiple machine learning models are multiple instances of a same machine learning model;

wherein the data indicates measured values from one or more engines;

wherein the selected portion of the data is provided by a selection by one or more users via one or more interfaces.