基于凸集模型的船舶坐墩配墩鲁棒设计方法

提出了船舶坐墩支墩配置的鲁棒设计方法.用凸集模型描述支墩组合刚度的不确定性,用极小化不确定性参数不满意度函数的方法实现目标函数的鲁棒性,用一个优化过程分析约束函数的最劣值实现约束函数的鲁棒性.计算示例表明,用所提方法得到的设计方案优于直接极大化不确定性参数最小值的结果.

Robust Design of Docking Blocks Using Convex Models

The positioning and stiffness allocation of docking blocks are very important when docking a ship, because mis-positioning or mis-allocation of docking blocks may give rise to unreasonably large block reactions and consequently serious damage to both the docked ship and the blocks. Various types of wood are used in the construction of docking blocks. The elastic modulus of various types of wood varies greatly depending on its age, moisture content, and effects of previous loading on its proportional limit. All these factors lead to the uncertain block stiffness and hence the uncertain equivalent block stiffness, which encompasses stiffness of a docking block and that of its supporting structure. Therefore, docking block stiffness may be treated as an uncertain variable in the design of docking blocks. To describe the uncertainty of equivalent block stiffness, the proposed method employs the convex model in which the indeterminacy about the uncertain variables is expressed in terms of some sets to which these uncertain variables belong. In this paper, a method of robust design with respect to the uncertain equvalent stiffnesses of blocks is proposed using the convex model with the help of the unsatisfactory degree functions. Robust design aims to achieve a state of robustness so that the performance of a design is least sensitive to the variability of uncertain variables. The robustness of the objective function is achieved by minimizing the worst value of unsatisfactory degree functions of the uncertainty parameters while the feasibility robustness is ensured by an optimisation conducting the worst-case analysis at a lower level. The proposed model is a nested optimization and a decomposition technique is then employed to relieve computational burdens. Numerical simulations show that the optimal result from minimization of the worst value of unsatisfactory degree functions is superior to that from direct maximization of the worst value of uncertainty parameters.