Screening analysis to detect adulteration in diesel/biodiesel blends using near infrared spectrometry and multivariate classification

This paper proposes an analytical method to detect adulteration of diesel/biodiesel blends based on near infrared (NIR) spectrometry and supervised pattern recognition methods. For this purpose, partial least squares discriminant analysis (PLS-DA) and linear discriminant analysis (LDA) coupled with...

Full description

Saved in:
Bibliographic Details
Published in:Talanta (Oxford) 2011-09, Vol.85 (4), p.2159-2165
Main Authors: Pontes, Márcio José Coelho, Pereira, Claudete Fernandes, Pimentel, Maria Fernanda, Vasconcelos, Fernanda Vera Cruz, Silva, Alinne Girlaine Brito
Format: Article
Language:English
Subjects:
Analytical chemistry
Biodiesel
Biofuels - analysis
Blends
Chemistry
Classification
Diesel
Diesel fuels
Diesel/biodiesel blends
Discriminant Analysis
Exact sciences and technology
Gasoline - analysis
Indexing in process
Least-Squares Analysis
Linear discriminant analysis
Multivariate Analysis
Near infrared spectrometry
Partial least squares-discriminant analysis
Pattern Recognition, Automated
Plant Oils - chemistry
Polymer blends
Principal Component Analysis
Raw
Screening analysis
Spectrometric and optical methods
Spectrophotometry, Infrared - methods
Successive projections algorithm
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:This paper proposes an analytical method to detect adulteration of diesel/biodiesel blends based on near infrared (NIR) spectrometry and supervised pattern recognition methods. For this purpose, partial least squares discriminant analysis (PLS-DA) and linear discriminant analysis (LDA) coupled with the successive projections algorithm (SPA) have been employed to build screening models using three different optical paths and the following spectra ranges: 1.0mm (8814–3799cm−1), 10mm (11,329–5944cm−1 and 5531–4490cm−1) and 20mm (11,688–5952cm−1 and 5381–4679cm−1). The method is validated in a case study involving the classification of 140 diesel/biodiesel blend samples, which were divided into four different classes, namely: diesel free of biodiesel and raw vegetal oil (D), blends containing diesel, biodiesel and raw oils (OBD), blends of diesel and raw oils (OD), and blends containing a fraction of 5% (v/v) of biodiesel in diesel (B5). LDA-SPA models were found to be the best method to classify the spectral data obtained with optical paths of 1.0 and 20mm. Otherwise, PLS-DA shows the best results for classification of 10mm cell data, which achieved a correct prediction rate of 100% in the test set.
ISSN:0039-9140
1873-3573
DOI:10.1016/j.talanta.2011.07.064