US 12,339,933 B2
Method and system for getting desired property of blend by optimizing blending rules
Vishnu Swaroopji Masampally, Pune (IN); Aditya Pareek, Pune (IN); and Venkataramana Runkana, Pune (IN)
Assigned to TATA CONSULTANCY SERVICES LIMITED, Mumbai (IN)
Filed by Tata Consultancy Services Limited, Mumbai (IN)
Filed on Dec. 13, 2022, as Appl. No. 18/080,054.
Claims priority of application No. 202121058609 (IN), filed on Dec. 16, 2021.
Prior Publication US 2023/0195853 A1, Jun. 22, 2023
Int. Cl. G05B 13/04 (2006.01); G06F 18/10 (2023.01); G06F 18/24 (2023.01)
CPC G06F 18/24765 (2023.01) [G05B 13/042 (2013.01); G06F 18/10 (2023.01)] 16 Claims
OG exemplary drawing
 
1. A processor implemented method for getting a desired property of a blend by optimizing a set of blending rules, the method comprising:
collecting, via one or more hardware processors, data from a plurality of data sources;
extracting, via the one or more hardware processors, a relevant data from the collected data, wherein the relevant data comprises one or more of spectral data, quantitative structure data, chromatography data, functional group contribution data and a user defined data;
preprocessing, via the one or more hardware processors, the extracted relevant data, wherein the preprocessed data comprises clean data of a plurality of features, wherein the plurality of features are corresponding to the blend;
selecting, via the one or more hardware processors, a set of features from the plurality of features using a plurality of feature selection techniques;
creating, via the one or more hardware processors, a plurality of soft-sensors for the desired blend property using the preprocessed data of the selected set of features, wherein the plurality of soft-sensors is used to predict the desired property of a plurality of components used for blending;
selecting, via the one or more hardware processors, a soft-sensor from amongst the plurality of soft-sensors for the desired property of the plurality of components used for blending based on accuracy;
providing, via the one or more hardware processors, the set of blending rules for each feature of the selected set of features;
determining, via the one or more hardware processors, best blending rule from amongst the set of blending rules corresponding to each feature of the selected set of features, wherein the best blending rule is obtained from data of components used for blending such that the selected soft-sensor can be used to predict the desired blend property of mixture of components, wherein the best blending rule is determined using an optimization technique, the optimization technique comprises:
predicting the blend property for each blending rule of the set of blending rules for each feature of the soft-sensor used to predict the blend property of the components used in the blend,
comparing the predicted blend property with the desired blend property, and
minimizing an error in prediction of the blend property to select the best blending rule; and
building, via the one or more hardware processors, the soft-sensor for the desired blend property using the best blending rule.