Resistance Optimization Of Multi-navigational Ship Based On Improved Neural Network &SBD Technology

Multi-navigational ship is a new concept ship with high speed on surface and concealment underwater.According to this characteristic,ensuring excellent rapidity for both surface and underwater is a critical research topic.In order to improve the optimization ability of this ship,a SBD model based on improved neural network was proposed.Based on FFD method,this ship was reconstructed geometrically.A set of hull shape transformation method suitable for multi-navigational ship was constructed and ship shape samples were obtained.In order to verify the accuracy of numerical simulation,the surface resistance of DTMB5414 and underwater resistance of Suboff-AFF8 were numerically verified based on RANS method.In order to obtain a grid that takes into account the accuracy and calculation cost,the grid independence verification was carried out.Finally,the numerical simulation of the ship shape samples was carried out.The classical genetic algorithm was improved and four high-dimensional test functions were selected to test the performance of improved genetic algorithm.It was proved that the improved genetic algorithm could overcome the problem of falling into the local optimal solution.Based on general BP neural network,the approximate model of resistance was constructed.The training results showed that the prediction accuracy of general BP neural network is low.In order to solve the problem that general BP neural network is prone to fall into local extreme value and overfit,the improved genetic algorithm was introduced to optimize the initial weight threshold of BP neural network.Compared with general BP neural network,the results showed that optimized BP neural network has higher prediction accuracy in verification set and low resistance sample.Based on the improved resistance approximation model,NSGA-Ⅱ algorithm was used to optimize the surface and semi-submersible resistance of multi-navigational ships.According to the optimization results,three Pareto optimal solutions Opt1,Opt5 and Opt22 were selected,among which Opt1 has maximum surface resistance drop(4.4%)and Opt22 has maximum semi-submersible resistance drop(3.4%).The profiles of three optimal solutions were compared,and improved suggestions for the shapes of multi-navigational ship were put forward.The three optimal solutions were compared with mother ship in waveforms,and the predicted value was campared with numerical result to verify the reliability of the optimization results.