Data Fusion Based On Deep Learning And Its Application For FPSO Monitoring System

Since Chinese waters are rich in oil and gas, FPSO has been widely used indeveloping oil and gas field in the Bohai Sea and the Nanhai Sea. Moreover, astechnology advances and oil production increases continuously, FPSO-tonnageincreases constantly as well. In recent years, with the increase in FPSO-tonnage andcomplexity of marine environment, a number of FPSO mooring accidents happened.So this thesis is focused on the study that how the more cutting-edge, deep learningalgorithm can be applied to the optimization of data fusion technology and that howthe technology can be applied to predict the function of FPSO monitoring and earlywarning systems. And treating monitoring data of offshore oil vessel113as learningobjects, our study will verify the applied dominance of data fusion based on deeplearning in FPSO monitoring systems.At first, this paper describes the origin, background and related algorithms ofdeep learning, focusing on Deep Belief Networks algorithm in deep learning. Then, itgives a brief introduction to traditional data fusion technology and application ofartificial neural networks in the data fusion. But due to the limitations of artificialneural networks and its own defects, the efficiency and complexity of data integrationneed improving. Next, the paper introduces the FPSO monitoring and early warningsystems and taking offshore oil vessel113an example, analyzes the structure ofmonitoring and early warning systems.Subsequently, the thesis will analyze how data fusion based on Deep BeliefNetworks is applied to monitoring and warning systems, explain algorithm steps tobuild the application model and spells out the application mode. Finally, taking thedata actual working conditions as a learning object, we will implement the model themodel and analyze results, thus to verify the superiority of data fusion based on thedepth learning in early warning functions of FPSO.