A Spectral AutoML approach for industrial soft sensor development: Validation in an oil refinery plant. (July 2021)

• Record Type: Journal Article
• Authors: de Souza, Daniela C.M.; Cabrita, Luís; Galinha, Cláudia F.; Rato, Tiago J.; Reis, Marco S.
• Is Part Of: Computers & chemical engineering. Volume 150(2021)
• Publication Date: 2021-07
• Subjects: Diesel -- Property prediction -- Soft sensors -- Spectral pre-processing -- Infrared Spectroscopy -- Waveband selection; Chemical engineering -- Data processing -- Periodicals; 660.0285
• DOI: 10.1016/j.compchemeng.2021.107324
• Languages: English
• ISSNs: 0098-1354

Abstract

Highlights: Spectral AutoML enables fast development of robust inferential models from spectral data. It considers the combined effect of pre-processing, hyper-parameter tuning, band selection, model estimation and resolution definition. Spectral AutoML was compared with models developed following the classical development pipeline. The industrial test data set consists of 12 properties to be predicted from FTIR spectra. Spectral AutoML led to clearly better models for 8 out of 12 properties and minor improvements for 3 properties. Abstract: Spectral AutoML is a platform for fast development of PAT soft sensors that considers the combined effect of pre-processing, band selection, band-wise resolution definition, hyper-parameter tuning and model estimation. Spectral AutoML was compared with models developed under the classic paradigm, and their performance assessed on an independent test set. The validation study regards the prediction of 12 different diesel fuels properties, using FTIR-ATR spectra. The proposed framework led to clearly better predictions in 8 out of the 12 properties, and minor improvements in 3 properties. The Spectral AutoML results were obtained overnight, without interfering in the daily work of the users, while the benchmark models resulted from several months of work and fine tuning of the methods. The results demonstrated the added value of the proposed Spectral AutoML approach in terms of prediction accuracy, development time of the models and reduced dependence on resident experts.