The Optimization Study Of Integrated Design Of Torpedo

Torpedo is one of the most important underwater weapons in the current war.The performance in actual application is directly determined by the solution of integrated design of torpedo.The System of torpedo is composed of many subsystems,such as Control,Auto Navigation,Dynamic,Wire Guidance,Fuse,Power Supply,Combat,and Propelling,between each subsystem connected by coupling mode.The study of integrated design of torpedo is full of difficulty due to the complicated system.In the previous optimization study of integrated design of torpedo,overall torpedo system is usually spited into independent subsystems,and then do design optimization of each subsystems separately.This method reduced the difficulty of optimization effectively,but the optimization results are not always the optimal results of torpedo overall optimization.In recent years,the integrated design of torpedo is gradually becoming a new research interest with the help of Computer Simulation Technology and the development of torpedo integrated simulation techniques.In this thesis,we will focus on the optimization study of integrated design of torpedo based on the trajectory torpedo simulation.However,since the torpedo simulation system is consist of torpedo kinematics dynamic model,auto-navigation detection probability model,ballistic trajectory model,and target operation model,it’s really a big challenge to do design optimization based on a system with high dimension,strong-nonlinearity,nonformulation,non-smooth,long computation time cost,etc.In order to solve this difficult problem,here we will do the optimization study of integrated design of torpedo as follows:(1)First,the problem of torpedo simulation model we are facing to is a black-box problem of seven dimensions,strong-nonlinearity,non-formulation,non-smooth,long computation time cost.In order to solve this complex optimization problem,we choose Genetic Algorithm to do the optimization.Genetic Algorithm has the advantages that it’s unconstrained by function derivative and continuous,only uses the value of the function itself for optimization,can quickly converge to the global approximate optimal solution,and can easily promote to multi target.When the population size is 10,iterations are 20,and the initial optimal individual is 0.41,after 20 hours of calculation,the result is 0.3455.We finally got a practical approved result.At the same time,we directly apply Genetic Algorithm to the multi target optimization of the overall multidisciplinary design model of torpedo.After 21 hours of calculation,we got two set of Pareto solutions.(2)Furthermore,in order to overcome the huge time cost of the model simulation,we applied Space Mapping Algorithm to the model simulation of the overall design optimization of torpedo.Through the building of coarse model,we transferred the work of torpedo simulation in searching to the fast coarse model.In order to build a better coarse model,we first analyzed the overall multidisciplinary design model of torpedo,and did the dimension reduction process through sampling.Then we revealed the distribution of extreme points with the sample repetitive rate.Last,we built a fine coarse model using Neural Network Algorithm through sample improvement by spline interpolation and analyzation of the model.(3)In the end,we optimized the overall multidisciplinary design model of torpedo using Space Mapping Algorithm based on the fine coarse model.The time of the whole optimization is only 24 minutes and the result of the optimization is 0.3436.Compared to costing 20 hours using Genetic Algorithm and the optimization result 0.3455,we can say,using Space Mapping Algorithm,we can get a good optimization result in a very short time.