Research On General Design Optimization Method Of Human Occupied Vehicle

Aiming at the disadvantage of traditional Human Occupied Vehicle(HOV)serial design mode,in this paper,the National Science and Technology Major Special Sub Task "Research on Deep Light Load Operation HOV Concept Design " and the Ministry of Education Scientific Research Project "Research on HOV Comprehensive Optimization" are taken as the background,the Multidisciplinary Design Optimization(MDO)is introduced into the Deep Light load operation HOV(DLHOV)general design,the main work is as follows:System decomposition is carried out on the base of considering the interaction among systems.The DLHOV is divided into eight disciplines,which are resistance,thrust,structure,guide and control,hydraulic operation,trim and ballast,energy,life support and communications.Design parameters of DLHOV general design are determined.Date exchanging relationships between system layer and disciplines、disciplines and disciplines respectively are cleared.Test comparisons among three approximation models(response surface,RBF model,Kriging model)is carried out.And then,the characteristics and applicability of these three approximation models are summarized.In addition,the discipline analysis models of structure discipline and resistance discipline are established by approximate model,and other disciplinary analysis models are established by formula estimating method.The general arrangement,floating state,equipment interference are considered as restricted conditions for system layer and brought into general design model.Test work on three single objective searching strategies(Nonlinear Programming Quadratic Line search,Multi-Island Genetic Algorithm,Algorithm of Simulated Annealing)is carried out,after which the above three single objective searching strategies are compared aiming at the abilities of obtaining global explain and calculation cost.Additionally,the research on hybrid methods is carried out.Test work on four Multi-objective searching strategies(Neighborhood Cultivation Genetic Algorithm,Fast and Elitist non-dominated sorting Genetic Algorithm,Multi-objective Simulated Annealing Algorithm,Multi-objective Particle Swarm Optimizations)is carried out,after which the above four Multi-objective searching strategies are compared aiming at distribution of Pareto solution set,the test results show that Fast and Elitist non-dominated sorting Genetic Algorithm(NSGA-II)can get satisfactory Pareto solution set on the basis of reasonable setting of algebra and population.The improved two-stage optimization strategies of Collaborative Optimization,Concurrent Subspace Optimization,Bi-level Integrated System Synthesizes are compared,and the test results show that Concurrent Subspace Design(CSD)has strong robustness in the process of searching,getting the global optimal solution quickly and proposing the CSD-(NSGA-II)MDO-MOP framework.In this paper,CSD is combined with DLHOV general design optimization model,and the DLHOV general design optimization framework is proposed.Then the framework is calculated,and the DLHOV general design scheme can be realized.