Publicação
Calibration transfer between benchtop and handheld near-infrared instruments for predictive models of commercial milk powder samples using cubic spline interpolation, piecewise direct standardization and artificial neural networks
| dc.contributor.author | Fhuaráin, Áine M. Ní | |
| dc.contributor.author | Guedes, Gonçalo M. | |
| dc.contributor.author | Zhao, Ming | |
| dc.contributor.author | Ferreira, António C. S. | |
| dc.contributor.author | O'Donnell, Colm P. | |
| dc.contributor.author | Gowen, Aoife A. | |
| dc.date.accessioned | 2026-07-29T16:33:22Z | |
| dc.date.available | 2026-07-29T16:33:22Z | |
| dc.date.issued | 2026-10-15 | |
| dc.description.abstract | Milk powder manufacture extends the shelf-life of milk which facilitates an increase in export opportunities, reduces storage costs and provides an ingredient for food applications. Using Near-Infrared (NIR) spectroscopy predictive models to predict compositional properties of milk powders is important for process and quality control applications. There is a need for calibration transfer, as new instruments are commissioned, including inexpensive, portable handheld instruments. Calibration transfer from benchtop to inexpensive, compact handheld instruments would increase quality measurements without the costly and time-consuming process of validating new calibration models using wet chemistry reference data. In this paper, NIR spectra of 70 samples of Full Fat Milk Powder (FMP) and 59 samples of Skim Milk Powder (SMP) were acquired on a benchtop and two handheld instruments of the same model and specification. Partial Least Squares (PLS) regression and Partial Least Squares-Discriminant Analysis (PLS-DA) were used to develop regression and classification models. Cubic spline interpolation, Piecewise Direct Standardization (PDS) and Artificial Neural Networks (ANNs) were used to perform calibration transfer on spectra from the handheld instrument to the benchtop instrument with the range cropped to match the handheld instrument. The second handheld instrument with the same specifications was then used to take spectra of a subset of the powders to investigate if the calibration transfer models were robust when used with independent instruments of the same type. ANNs yielded a lower Root Mean Square Error Predictions (RMSEP) than PDS for the independent handheld instrument, when the transferred spectra are inputted into the master models. | eng |
| dc.identifier.doi | 10.1016/j.chemolab.2026.105809 | |
| dc.identifier.eid | 105045170835 | |
| dc.identifier.other | c9872b58-61eb-4729-b58c-a13d29d5f30a | |
| dc.identifier.uri | http://hdl.handle.net/10400.14/58913 | |
| dc.identifier.wos | 001827028900001 | |
| dc.language.iso | eng | |
| dc.peerreviewed | yes | |
| dc.publisher | Elsevier B.V. | |
| dc.rights.uri | http://creativecommons.org/licenses/by/4.0/ | |
| dc.title | Calibration transfer between benchtop and handheld near-infrared instruments for predictive models of commercial milk powder samples using cubic spline interpolation, piecewise direct standardization and artificial neural networks | |
| dc.type | research article | |
| dspace.entity.type | Publication | |
| oaire.citation.volume | 277 | |
| oaire.version | http://purl.org/coar/version/c_970fb48d4fbd8a85 |
Ficheiros
Principais
1 - 1 de 1