US 12,346,071 B2
Method and control unit for automatically selecting data sets for a method for machine learning
Simon Weissenmayer, Flein (DE); and Daniel Stuemke, Backnang (DE)
Assigned to ROBERT BOSCH GMBH, Stuttgart (DE)
Appl. No. 17/768,353
Filed by Robert Bosch GmbH, Stuttgart (DE)
PCT Filed Oct. 8, 2020, PCT No. PCT/EP2020/078224
§ 371(c)(1), (2) Date Apr. 12, 2022,
PCT Pub. No. WO2021/073995, PCT Pub. Date Apr. 22, 2021.
Claims priority of application No. 10 2019 216 025.5 (DE), filed on Oct. 17, 2019.
Prior Publication US 2024/0142916 A1, May 2, 2024
Int. Cl. G05B 13/02 (2006.01); G01D 3/02 (2006.01); G06F 3/06 (2006.01)
CPC G05B 13/0265 (2013.01) [G01D 3/022 (2013.01); G06F 3/0608 (2013.01); G06F 3/0652 (2013.01)] 11 Claims
OG exemplary drawing
 
1. A method for automated selection of data sets for a method for machine learning for detecting operating variables of a motor vehicle, the method for the automated selection comprising the following steps:
detecting, during operation of the motor vehicle, measuring signal sequences for a plurality of operating variable ranges of the motor vehicle;
assigning the operating variable ranges memory areas, each including multiple slots which are each configured to store a data set containing a detected measuring signal sequence of the detected measuring signal sequences, wherein data sets already stored in the slots are overwritable with data sets newly detected instantaneously for the same memory area; and
forming an estimated slot error value of the detected measuring signal sequence for each of the slots of each memory area in which a data set is stored, and storing the estimated slot error value together with the detected measuring signal sequence;
wherein, those of the data sets whose estimated slot error values are lower than further estimated slot error values that have been formed for the same memory area are overwritten.