Incremental Learning in Modelling Process Analysis Technology (PAT)—An Important Tool in the Measuring and Control Circuit on the Way to the Smart Factory

To meet the demands of the chemical and pharmaceutical process industry for a combination of high measurement accuracy, product selectivity, and low cost of ownership, the existing measurement and evaluation methods have to be further developed. This paper demonstrates the attempt to combine future...

Full description

Saved in:
Bibliographic Details
Published inSensors (Basel, Switzerland) Vol. 21; no. 9; p. 3144
Main Authors Choudhary, Shivani, Herdt, Deborah, Spoor, Erik, García Molina, José Fernando, Nachtmann, Marcel, Rädle, Matthias
Format Journal Article
LanguageEnglish
Published Basel MDPI AG 01.05.2021
MDPI
Subjects
Accuracy
Algorithms
Chemical industry
Communication
Datasets
Energy
Ethanol
Evaluation
Hypotheses
incremental learning
Laboratories
Learning
Measurement methods
Monitoring
Photometers
Process controls
process technology
Profits
Raman spectroscopy
Selectivity
Spectrum analysis
Support vector machines
SVM
Technology assessment
Germany
Online AccessGet full text

Cover

More Information
Summary:To meet the demands of the chemical and pharmaceutical process industry for a combination of high measurement accuracy, product selectivity, and low cost of ownership, the existing measurement and evaluation methods have to be further developed. This paper demonstrates the attempt to combine future Raman photometers with promising evaluation methods. As part of the investigations presented here, a new and easy-to-use evaluation method based on a self-learning algorithm is presented. This method can be applied to various measurement methods and is carried out here using an example of a Raman spectrometer system and an alcohol-water mixture as demonstration fluid. The spectra’s chosen bands can be later transformed to low priced and even more robust Raman photometers. The evaluation method gives more precise results than the evaluation through classical methods like one primarily used in the software package Unscrambler. This technique increases the accuracy of detection and proves the concept of Raman process monitoring for determining concentrations. In the example of alcohol/water, the computation time is less, and it can be applied to continuous column monitoring.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 23
These authors contributed equally to this work.
ISSN:1424-8220
1424-8220
DOI:10.3390/s21093144