Font Size: 

Reliability-baseddesign optimization (RBDO) is regarded as a reasonable and powerful tool forstructural optimization because of rational consideration of uncertainties. Probabilisticconstraints evaluation in RBDO can be carried out using theperformance measure approach (PMA) other than the traditional reliability indexapproach (RIA). In PMA, the probabilistic performance measure (PPM) is obtainedthrough locating the minimum performance target point (MPTP) with the specifiedtarget reliability index in standard normal space, which is also called inversereliability analysis. The advanced mean-value (AMV) method is well suitable forlocating MPTP due to its simplicity and efficiency. However, AMVmay converge very slowly, or oscillate and fail to converge if the performancefunction is concave and highly nonlinear. A step length adjustment (SLA) iterativealgorithm, which introduced a “new” step length to control the convergence ofthe sequence, has been proposed by the authors. This step length is new becausethe line search process for step length selection is not needed and it may beconstant during the whole iteration process or decrease successively severaltimes using a self-adjust strategy. SLA is as simple as AMV and does not needthe prior knowledge of convexity or concavity of the performance function asother modified algorithms do. It has been proved that the AMV method is aspecial case of the SLA algorithm when the step length tends to infinity.Inthis paper, several deliberately designed numerical examples are used tocompare SLA with AMV and other improved algorithms, including hybrid mean valuemethod (HMV), chaos control (CC) method and modified chaos control (MCC)method. The results indicate that SLA is effective and robust. In RBDO, thedesign variables can be parameters of the probability distribution of therandom variables, such as mean values. During the optimization, designvariables change,i.e., mean values vary in the design space. One numerical example show that when the mean value vary, AMV cannot obtainconvergent solution for PPM during  somevariation intervals, which will lead to failure in the optimization.However, SLA acquires the stable convergence solution for the whole variationinterval. Then SLA is used to compute PPM and RBDO is executed. Several examples illustratedthat RBDO with SLA is also effective and more robust thanother algorithms.