US 11,809,987 B2
Computer-implemented methods for training a machine learning algorithm
Christopher R Hall, Derby (GB); Malcolm L Hillel, Derby (GB); Bryce D Conduit, Derby (GB); Anthony M Dickens, Cambridge (GB); James V Taylor, Cambridge (GB); and Robert J Miller, Cambridge (GB)
Assigned to ROLLS-ROYCE plc, London (GB)
Filed by ROLLS-ROYCE plc, London (GB)
Filed on Jun. 10, 2020, as Appl. No. 16/897,903.
Claims priority of application No. 1908494 (GB), filed on Jun. 13, 2019.
Prior Publication US 2020/0394517 A1, Dec. 17, 2020
Int. Cl. G06N 3/08 (2023.01); G06N 3/02 (2006.01); G06N 3/045 (2023.01); G05B 23/02 (2006.01); G06N 7/04 (2006.01)
CPC G06N 3/08 (2013.01) [G05B 23/024 (2013.01); G06N 3/02 (2013.01); G06N 3/045 (2023.01); G06N 7/046 (2013.01)] 17 Claims
OG exemplary drawing
 
1. A computer-implemented method comprising:
controlling input of at least a portion of a first training data set that is input to a first machine learning algorithm to train the first machine learning algorithm, the first training data set including: data quantifying damage to a first compressor; and data quantifying a measured first operating parameter of the first compressor, wherein generating the data quantifying damage to the first compressor comprises:
receiving data quantifying damage received by one or more compressor blades of the first compressor,
using the received data quantifying damage received by one or more compressor blades of the first compressor and a plurality of damage parameters, generating the data quantifying damage to the first compressor, and
determining importance of at least a subset of the plurality of damage parameters, and where a first damage parameter has an importance that does not meet a predetermined criterion, re-generating the data quantifying damage to the first compressor without using the first damage parameter;
receiving data quantifying a predicted first operating parameter as output from the first machine learning algorithm based on the data quantifying damage to the first compressor that is input as the first training data set; and
training the first machine learning algorithm by comparing the received data quantifying the predicted first operating parameter as output from the first machine learning algorithm to the data quantifying the measured first operating parameter of the first compressor that is input as the first training data set, the trained first machine learning algorithm being configured to enable determination of operability of a second compressor of a gas turbine engine.