Multi-objective Optimization Method In The Transverse Section Of Midship Based On Robustness

In traditional structure optimization, the Optimization solution obtained may usually be at the edge of the feasible region without the consideration about the influence from the uncertainties of design variables and others. Once some fluctuation occurs in design variables, the optimization solution may deviate from the feasible region and becomes non-feasible solution. Robust design is a kind of method whose aim is to reduce the sensitivity of quality of the products to controllable factors and uncontrollable factors by adjusting the size of the product design variables. After robust design, the product quality can still meet the design requirements when controllable factors and uncontrollable factors change.Robust design have been applied in hull structure optimization, but in some of these applications, hull structure characteristics and the load responses often are usually obtained by empirical formulas. However, because of the complexities of the structures and applied load of hull structure, the structural responses and interactions among loads tend to become very complicated, if the structure characteristic and the load responses are just expressed by empirical formulas simply, the interactions among structures and loads will be neglected, result in a large deviation in optimization result. So it is necessary to carry out Finite Element Analysis Method of hull structure to accurately analyze the structural characteristics and structural responses, then make these results as objective function or constraints characteristic values the ship’s structure optimization objective function values and constraints characteristic values. In this paper, combine the idea of robust optimization with finite element analysis, a kind of robust optimization method is proposed based on finite element analysis results. For the purpose, firstly, describe the basic theory of robustness, principles, objectives, design principles and the corresponding design flow of robust design,introduce all kinds of traditional robust design method. Based on the above, a kind of multi-objective particle swarm optimization algorithm for robust optimal solution by matlab software. Then make particle swarm optimization algorithm written by Matlab associated with finite element software MSC Patran, complete the single objective optimization whose objectives are respectively weight and maximum stress in cabin structure of multi-purpose vessel. Next in order to improve the availability of robust design method proposed in hull structure optimization, the support vector machine(SVM) approximate model is chosen to replace finite element analysis, for the purpose to improve performance of SVM and the accuracy of approximate model, the optimization for important parameters of SVM is carried out.Finally, By using the method of robust design proposed in this paper and the SVM approximate model, complete the robust design of transverse section of midship, then make the comparison between the robust optimization results and traditional optimization results.From the optimization results, we can know that the robustness of cabin become better, the maximum stress of the cabin and its sensitivity to the design variables have reduced to a greater degree. At the same time, SVM approximate model have made the time consumption of optimization procedure reduced greatly, improve the availability of method proposed.