Recognition Method of Petroleum Fluorescence Spectra Based on Convolutional Neural Network
Abstract As an important raw material and petrochemical tool, petroleum not only brings convenience to mankind, but also brings huge socio-economic and cultural value to our social development, but at the same time it also causes a lot of serious damage to our ecological environment. The identification and measurement of petroleum pollutants has become the main tool to identify pollution sources, control their pollutants and protect their ecological environment. This paper explores the petroleum fluorescence spectrum identification method based on convolutional neural network. Based on extensive research on this method, a simple analysis and understanding of petroleum fluorescence spectrum identification technology and petroleum-related principles are carried out, and then summarized according to relevant data find out the main factors that affect fluorescence spectrum recognition, and prepare for the experiment. The feasibility of the method is verified through the petroleum fluorescence spectrum recognition experiment of the convolutional neural network. The experimental results show that the relative error of the fluorescence spectrum recognition under different concentrations of petroleum both are within the range of 9%. Through the analysis of the relative error, it can be seen that the relative error of resolution shows a downward trend with the increase of the concentration. According to the above data, it can be seen that when the convolutional neural network algorithm is used to identify the components of the petroleum mixed solution, the qualitative analysis can be completed well. When the components in the mixed solution are quantitatively analyzed, there is a certain relative error.