Oil source recognition technology using concentration-synchronous-matrix-fluorescence spectroscopy combined with 2D...

Area of Science:

• Analytical Chemistry
• Spectroscopy
• Machine Learning

Background:

• Accurate oil source recognition is crucial for protecting water resources from pollution.
• Existing methods may lack the speed, accuracy, or economic viability required for real-time applications.
• Environmental factors like weathering and seawater contamination complicate oil identification.

Purpose of the Study:

• To develop a rapid, accurate, and economical method for recognizing the source of crude oil and petroleum products.
• To evaluate the performance of different pattern recognition algorithms for oil identification.
• To assess the effectiveness of the proposed method under realistic environmental conditions.

Main Methods:

• Concentration-Synchronous-Matrix-Fluorescence (CSMF) spectroscopy was employed to obtain spectral data.
• Two-dimensional (2D) wavelet packet analysis was used to extract feature vectors from CSMF contour images.
• Four pattern recognition algorithms were compared: Back-propagation (BP) neural network, Radial Based Function Neural Network (RBFNN), Support Vector Machine (SVM), and Probabilistic Neural Network (PNN).
• Oil samples were subjected to weathering and seawater adulteration to simulate real-world contamination scenarios.

Main Results:

• The Probabilistic Neural Network (PNN) demonstrated the highest recognition accuracy among the tested algorithms.
• PNN also exhibited the fastest recognition time, making it suitable for rapid analysis.
• The CSMF spectroscopy combined with 2D wavelet packet feature extraction and PNN proved effective even with interference factors.
• The method showed potential for identifying unconventional oil types.

Conclusions:

• The proposed method integrating CSMF spectroscopy, 2D wavelet packet analysis, and PNN offers a superior approach for oil source recognition.
• PNN provides a robust and efficient solution for accurate and rapid identification of oil types in environmental samples.
• This technique holds promise for applications in oil spill monitoring and water resource protection, particularly for unconventional oil sources.