A Model For Deepwater Floating Platforms Selection Based On BP Artificial Neural Networks

Selection of floating platforms is influenced by many factors, such as water depth, risers, capacity of production, experience of operators, etc. Choosing floating platforms for a specific oil/gas field in deepwater must be based on the consideration of particular condition of that field. This paper investigates the application of floating platforms (including Spar, TLP, SEMI and FPSO) in deep water developments all around the world, and selects nine of the influencing factors to build a model for deepwater floating platforms selection by a method of BP (Back-propagation) Artificial Neural Networks(ANN)improved using the L-M algorithm. This prediction model was applied in West Africa oil/gas Fields Egina. The result shows that improved prediction model for deepwater floating platform selection using L-M algorithm has a high convergence speed and good accuracy. In addition, the prediction result of this model for Egina is similar to that of analysis according to experience. Also, analysis for causes of errors in mathematical model lays the groundwork for model optimizing.