Research On The Engineering Structural Robust Design Method Based On Stochastic Optimization Model

Structural design is subjected to uncertainties in material, geometrical properties,loads and other variables. Reliability design is an advanced and efficient approach to handle the uncertainties,and robust design is a new quality-oriented method used to improve robustness of structure. The performance of robust structure is not sensitive to the change of the design parameters and noises.The traditional robust design is a method based on the experiment. Its core is Taguchi's quality view, and its most essential step is parameter design and tolerance design. But the function and applicable scope of traditional analysis method, such as the variance analysis, have the certain limit,and the design procedure is tedious.At first, the core ideas of robustness is refined from the basic concepts,principle and procedure of Taguchi Method , a traditional method used early for robust design. In second, the variables, constrains and targets are discussed in the view of probability,optimization and robustness, moreover, a new robust design model based on probabilistic reliability theory and optimization technique is presented, in which the minimization of the objective function is constrained by the reliability requirements. The new model takes advantage of the tolerance model and the ordinary stochastic optimization model, may deal with not only the certain variables with tolerance or zero-tolerance ,but also the uncertain variables with randomness of design factors and their probability statistical parameters,so it is more practicable. At last,an approach and its programming procedure are derived to solve the problem with aid of MATLAB tools about GA(Genetic Algorithms),and the given examples show that the robust design is more advanced in technique and more necessary in practice than the reliability design of the popular design code.