The Research Of Parameters Optimization Of FPSO Hull Grider Structure Based On BP Neural Network And Fruit Fly Optimization Algorithm

With the development of exploitation of offshore oil,floating,production,storage and offloading(FPSO)unit has played a leading role for the production,processing and storage of oil,especially in deep water and in remote areas.The hull girder ultimate strength of FPSO,which is related to the safety and economic aspects,is usually designed and checked based on engineer's experience.Although many classification societies have their methods and software for the FPSO design,all the methods belong to enumeration method.We can get a better set of design parameters,but it is very difficult even impossible to find the global optimum of design parameters using these methods.In this paper,a variant of backpropagation neural network,named IFOA-BP neural network,is proposed based on an improved fruit fly optimization algorithm(IFOA).According to a limited amount of training data,the IFOA-BP model can predict the strain value of the critical position after giving any set of design parameters(in this study,four design parameters are adopted).Simulation results show that the predication performance of IFOA-BP neural network is better than that of standard BP model and GA-BP model.Then,IFOA which introduces an adaptive changing iteration step value,is employed to search for the optimal set of the four design parameters based on the selected best IFOA-BP model.Compared with the set of the four design parameters of an existing FPSO,the optimal set of design parameters obtained in our study is better in terms of the strain value of critical position and the self-weight of FPSO model.In summary,the main contribution of the study is that we provide a feasible and high-efficiency method to optimize the design parameters of FPSO.