Oil source recognition technology using concentration-synchronous-matrix-fluorescence spectroscopy combined with 2D wavelet packet and probabilistic neural network.

Developing an accurate, rapid and economic oil source recognition method is essential for water recourses protection. Concentration-synchronous-matrix-fluorescence (CSMF) spectroscopy combined with 2D wavelet packet and probabilistic neural network (PNN) was proposed for source recognition of crude oil and petroleum products samples in this study. 2D wavelet packet was used to extract wavelet packet coefficients as the feature vectors from CSMF contour image and four algorithms, Back-propagation (BP) neural network, Radial based function neural network (RBFNN), Support vector Machine (SVM) and probabilistic neural network (PNN) were carried out for pattern recognition. With the introduction of interference factors such as weathering and sea water adulteration to the three samples from Bohai bay territory of China, the comparison about accuracy and recognition time of the four methods was discussed and the results showed that PNN network maintain the highest recognition accuracy and speed. These findings may offer potential application for oil spill recognition for unconventional oil.