An efficient classification method for fuel and crude oil types based on m/z 256 mass chromatography by COW-PCA-LDA. (15th June 2018)

• Record Type: Journal Article
• Authors: Sun, Peiyan; Bao, Kaiwen; Li, Haoshuai; Li, Fujuan; Wang, Xinping; Cao, Lixin; Li, Guangmei; Zhou, Qing; Tang, Hongxia; Bao, Mutai
• Is Part Of: Fuel. Volume 222(2018)
• Page Start: 416
• Page End: 423
• Publication Date: 2018-06-15
• Subjects: Crude oil -- Fuel oil -- m/z 256 mass chromatography -- Correlation optimized warping (COW) -- Principal component analysis (PCA) -- Linear discriminant analysis (LDA); Fuel -- Periodicals; Coal -- Periodicals; Coal; Fuel; Periodicals; 662.6
• DOI: 10.1016/j.fuel.2018.02.150
• Languages: English
• ISSNs: 0016-2361

Abstract

Highlights: 300 oil samples were analyzed by GC–MS. The m / z 256 mass chromatogram was chosen for classification and identification. COW, PCA and LDA methods were combined to identify the type of oils and only one fuel sample was misjudged as crude oil. The good model with three discriminant equations for HFOs, LFOs and OILs was obtained. Abstract: Chemometric analysis was used to classify different oils based on the m / z 256 mass chromatogram. Three hundred oil samples comprised of 12 LFOs, 104 HFOs (13 weathered fuel oils), and 184 crude oil samples (63 weathered crude oils) were analyzed by GC–MS, and the m / z 256 mass chromatogram was chosen for classification. After normalization, the m / z 256 mass chromatograms were aligned using the correlation optimized warping (COW) method with segment lengths of 50 data points and slack parameters of 4 data points. They were then analyzed by principal component analysis (PCA). The score graphs of the PCs revealed that there was a good discriminant for HFOs, LHOs and oils. According to the changing tendency of the correct sample percentage, five PCs were chosen for the linear discriminant analysis (LDA). A good model with three discriminant equations for HFOs, LFOs and oils was obtained. For the training set with 172 samples, the correct percentage reached 100%, and 99% was obtained for the test set with 128 samples. This study proved that the COW-PCA-LDA method based on the m / z 256 mass chromatogram can be used in oil fingerprinting identification, especially for oil type classification.