US 12,147,222 B2
Computer-implemented determination of a quality indicator of a production batch-run of a production process
Reinhard Bauer, Heidelberg (DE); Martin Hollender, Dossenheim (DE); and Marco Gaertler, Dossenheim (DE)
Assigned to ABB Schweiz AG, Baden (CH)
Filed by ABB Schweiz AG, Baden (CH)
Filed on Oct. 14, 2021, as Appl. No. 17/500,982.
Application 17/500,982 is a continuation of application No. PCT/EP2020/060203, filed on Apr. 9, 2020.
Claims priority of application No. 19169994 (EP), filed on Apr. 17, 2019.
Prior Publication US 2022/0035351 A1, Feb. 3, 2022
Int. Cl. G05B 19/418 (2006.01)
CPC G05B 19/41875 (2013.01) [G05B 2219/32077 (2013.01)] 13 Claims
OG exemplary drawing
 
1. A computer-implemented method to determine a quality indicator (Q) of a particular production batch-run in accordance with a predefined production process, wherein technical equipment performs the predefined production process and thereby provides data in the form of time-series from a set of data sources, wherein the data sources are related to the technical equipment, the method comprising:
obtaining a conversion factor vector by applying a machine-learning algorithm to reference data from two reference batch-runs of the predefined production process, wherein the conversion factor vector comprises a set of source-specific factors for the set of data sources;
receiving a multi-variate time-series ({{R}}) from a further reference batch-run of the predefined production process, the further reference batch-run of the predefined production process conforming to a particular quality category, the multi-variate time-series ({{R}}) having a set of uni-variate time-series ({Rv}) from the set of data sources;
converting the multi-variate time-series ({{R}}) from the further reference batch-run into a converted reference time-series ({R} #) that is uni-variate, by multiplying data values of each of the set of uni-variate time-series ({Rv}) with a respective source-specific factor from the conversion factor vector, followed by summing up the multiplied data values of the set of uni-variate time-series ({Rv}) according to discrete time points of the multi-variate time-series ({{R}}) of the reference batch-run;
receiving a production multi-variate time-series ({{P}}) from the production batch-run of the predefined production process, the production multi-variate time-series ({{P}}) having a set of uni-variate time-series ({Pv}) from the set of data sources;
converting the production multi-variate time-series ({{P}}) from the production batch-run into a converted production time-series ({P} #), by multiplying data values of each of the uni-variate time-series ({Pv}) with the respective source-specific factor of the conversion factor vector, followed by summing up the multiplied data values of the set of uni-variate time-series (({Pv})) according to discrete time points of the multi-variate time-series ({{P}} from the production batch-run;
comparing the converted reference time-series ({R} #) with the converted production time-series ({P} #), thereby determining whether the production batch-run of the predefined production process conforms to a particular quality category or does not conform to the particular quality category; and
controlling parameter settings of the production batch-run based on whether the production batch-run of the predefined production process conforms to the particular quality category or does not conform to the particular quality category, wherein the parameter settings of the production batch-run relate to one or more components of the technical equipment that performs the predefined production process.