Laser-induced fluorescence detection of oil film thickness in water based on convolutional neural network algorithm

Laser-induced fluorescence spectroscopy plays an important role in rapid detection and identification of oil spills at sea. The fluorescence spectra of different thicknesses of oil films of the same type of oil are very similar, which makes it difficult for the method to detect and recognize the thickness of oil films. In this paper, a laser-induced fluorescence detection method based on convolutional neural network algorithm is proposed to measure the thickness of oil films. Experimental studies of laser-induced fluorescence were conducted for different oil spill film thicknesses, and fluorescence spectral data were obtained for oil-free water and for 10 different oil film thicknesses. The recognition of 11 samples with different oil film thicknesses was realized by Matlab software using convolutional neural network algorithm. It was verified that the method can completely recognize different thicknesses of oil film with high accuracy and low training set error. Based on the convolutional neural network algorithm, the oil spill volume can be calculated by detecting the oil spill thickness, which has certain application value for the protection of marine environment.