[Identification of Oil Type Using Spectral Reflectance Characteristics]

Area of Science:

• Spectroscopy
• Chemometrics
• Data Analysis

Background:

• Accurate identification of oil types is crucial for environmental monitoring and industrial applications.
• Traditional methods for oil identification can be time-consuming and labor-intensive.
• Developing rapid and precise methods for oil type discrimination is an ongoing research area.

Purpose of the Study:

• To investigate the feasibility of using spectral reflectance data for rapid oil type identification.
• To evaluate the effectiveness of cluster analysis and principal component analysis (PCA) in classifying different oil types.
• To determine the optimal analytical approach for distinguishing between kerosene, lubricating oil, and diesel fuels based on their spectral properties.

Main Methods:

• Analysis of reflectance spectra from four common oil types: kerosene, lubricating oil, light diesel oil, and 180# diesel, across various thicknesses.
• Application of cluster analysis with Euclidean distance to classify oil samples.
• Utilization of principal component analysis (PCA) on original data, wavelet approximate coefficients, and detail coefficients for feature extraction and classification.

Main Results:

• Cluster analysis achieved 100% classification accuracy at a specific cluster distance (L=8.976), demonstrating its effectiveness.
• Oil film thickness was found to influence the clustering results, highlighting its importance in spectral analysis.
• Principal component analysis showed that PCA scores of wavelet detail coefficients provided the best discrimination among oil types compared to other spectral components.

Conclusions:

• Spectral reflectance data, when combined with cluster analysis, provides a feasible method for oil type identification.
• Principal component analysis, particularly using wavelet detail coefficients, enhances the accuracy and efficiency of oil type classification.
• The proposed spectroscopic approach offers a fast, timely, and accurate solution for distinguishing between different oil types.