Parametric Sensitivity Analysis And Optimization For Nonrigid Airship

The capsule structure of nonrigid airship is a set of non-rigid system composed byskin membrane and cables. While the capsule gets aerated, the stiffness is formed. Inpractical work, the mechanical performance is mainly controlled by its materialproperties, and it is also highly related to its load case which brings uncertainty. Forthese2points, we should carefully study the influence level of the parameters to thestructural performance in this non-linear system (here defined as Parameter Sensitivity),and select those most important parameters for later optimization to achieve a desirableoptimal solution. In this text, I study on a nonrigid airship model to explore theapplication of the methods of sensitivity analysis and optimization by both StandardGenetic Algorithm and Pareto frontier based multi-objective genetic algorithm,furthermore, based on all these work I introduce a new strategy for improving themulti-objective genetic optimization. My work includes:1. By both local method and global method, sensitivity analysis is performed forevery desired parameter from material property, load cases and geometric informationunder different load cases such as vertical-bend, lateral-bend, axial-bend,air-pressure-filling and vibration. Local and global sensitivities are obtained andcompared with good similarity, and the most sensitive parameters are selected for lateroptimization. At last, an interpolation database of the nonrigid airships is realized fordirect parameter sensitivity inquiry in order to avoid massive calculation.2. A preliminary optimization for nonrigid airships is conducted by the method ofstandard genetic algorithm with a composite objective function defined as mass/stiffness.The feasibility of genetic algorithm is proved good.3. With the ideas of Pareto frontiers in multi-objective optimization, a universaloptimization software is developed for the solutions to multi-objective problem.Executed by this software, two models (a3-D triangular cross-section truss with106parameters and a capsule structure of nonrigid airships) are optimized with ideal results.4. Inspired by the above work, an improved algorithm SRCC-SPEA IntegratedStrategy is proposed in the light of the guiding role of the parameter sensitivity playedin determining un-neglected parameter for optimization. Spearman’s rank correlationcoefficient is adopted as a sensitivity service provider, and integrated to the modifiedPareto frontier based genetic algorithm. This improved strategy is assembled in theoptimization software. Two examples proved its feasibility and efficiency.