Oil Tanker Structure Optimization Research Based On Improved Particle Swarm Optimization Algorithm

With the development of ship structure towards large-scale and complicated,the number of ship structure optimization design variables increases,the degree of nonlinearity of constraints and objective functions increases,and ship structure optimization problems exhibit multi-peak,high-dimensional and highly nonlinear characteristics.The PSO-BP neural network surrogate,which combines the global exploration ability of particle swarm optimization(PSO)with the local development ability of BP neural network,can effectively reduce the number of finite element analysis and improve the efficiency of structural optimization under certain accuracy.However,the population diversity of PSO algorithm can not be well maintained in the process of iterative evolution,and the algorithm has the problem of premature convergence.In this paper,a stage mutation particle swarm optimization(SMPSO)algorithm is used to optimize the parameters of BP neural network,which is applied to the structural optimization of oil tankers and achieves good results.The main contents of this paper are as follows:(1)Established a cabin finite element model of the 107600 DWT oil tanker.Sensitivity analysis was carried out by ISIGHT software to obtain effective design variables for ship structure optimization.BP neural network training and testing sample data are obtained by orthogonal experiment.(2)Introduced the basic principles and parameter settings of BP neural network and PSO algorithm,and completes BP and PSO-BP neural network programming through MATLAB software.BP neural network and PSO-BP neural network are trained by training sample data,and their generalization ability is tested and compared by testing sample data.(3)The particle swarm optimization algorithm is improved by a staging mutation strategy.That is,the initial position of particles is mutated in the initial stage,and the population optimal solution is disturbed by chaotic strategy in the middle and later stages,which keeps the population diversity of PSO algorithm changing continuously.Through the same training and testing,the generalization ability of SMPSO-BP neural network is better and more stable.(4)The SMPSO-BP neural network and the constrained minimization function fmincon are used to optimize the structure of the oil tanker cabin,and the effect is obvious.Finite element analysis of the optimized cabin model shows that the indicators meet the requirements of the specification,which shows that it is feasible to apply SMPSO-BP neural network to ship structure optimization.