荧光
材料科学
内容(测量理论)
计算机科学
环境科学
石油工程
计算机视觉
光学
地质学
数学
物理
数学分析
作者
Yunrui Hu,Xiaoyu Chen,Xinyi Li,Deming Kong
标识
DOI:10.1088/1361-6501/ae01c5
摘要
Abstract Oil spills on the sea surface are emulsified under the influence of wind and waves, forming emulsified oil with complex physicochemical properties. Accurate measurement of oil content for emulsified oil is essential for environmental protection and pollution control. Laser-Induced Fluorescence (LIF) is a widely used and effective technique for oil spill detection and oil content measurement. However, significant challenges remain when relying solely on LIF for measuring the oil content of emulsified oil. The overlapping fluorescence spectra caused by complex components reduce measurement accuracy and stability. Additionally, the multiphase structure and intricate physicochemical characteristics interfere with LIF signals, further limiting its reliability. To overcome these limitations, a multimodal data fusion method was proposed to improve measurement accuracy in this study. In the experiment, spectral data and corresponding images of emulsified diesel oil with different oil contents were collected using a self-developed device combined LIF and visual imaging technology. The images were decomposed into four frequency bands using wavelet transform, and the image features (energy, entropy, contrast and uniformity) were extracted. The spectral data were corrected and processed, then fused with the extracted image features. Partial Least Squares Regression (PLSR) was employed to fit the relationship between oil content and fusion data. To verify the method's applicability, emulsified crude oil from real samples were also analyzed. The results demonstrated that the fitting accuracy of the fused data was higher than that of the spectral data or image features alone, for both emulsified diesel oil or emulsified crude oil. Consequently, the proposed method enables accurate measurement of the oil content in emulsified oils, providing an efficient and reliable technical approach for oil spill monitoring and pollution control.
科研通智能强力驱动
Strongly Powered by AbleSci AI