Optimization based on reliability and confidence interval design for the structural-acoustic system with interval probabilistic variables

• O-RCID is proposed for optimization of structural-acoustic problem with interval probability variables.
• CV-IPPM is proposed to calculate the confidence interval of sound pressure response under interval probability model.
• CV-IPPM is developed to evaluate the system reliability under interval probability model.
• The effectiveness and efficiency of O-RCID based on CV-IPPM are investigated.

In practice, the uncertain parameters whose unique distributions cannot be identified on the basis of limited data existed in the structural-acoustic system extensively. The uncertain parameters with limited data can be modeled as interval probabilistic variables whose distribution parameters are expressed as interval variables instead of unique values. For the optimization of the structural-acoustic system with interval probabilistic variables, a new optimization technique named as the optimization based on reliability and confidence interval design (O-RCID) is developed. O-RCID is a nested loop optimization whose computational cost during the evaluation of the confidence interval of sound pressure response and the system reliability is considerable. To improve the computational efficiency of O-RCID, a change-of-variable interval probabilistic perturbation method (CV-IPPM) which is based on the perturbation approach and change-of-variable technique is proposed to calculate the confidence interval of sound pressure response and the system reliability. By introducing CV-IPPM, the nested loop O-RCID is converted into an approximate single-loop one which can be efficiently solved. Numerical examples demonstrate the effectiveness and efficiency of O-RCID based on CV-IPPM for the optimization of the structural-acoustic system with interval probabilistic variables.

As is shown in Fig. 1, the nested loop O-RCID under the interval probabilistic model is converted into an approximate single-loop one by employing CV-IPPM. Thus, the computational efficiency of the proposed O-RCID based on CV-IPPM is improved greatly, when compared with the nested loop O-RCID in which C¯α(Uj(d,X(x)))upper and P̲(gr(X(x))≥0) are estimated by the Monte-Carlo method.Figure optionsDownload as PowerPoint slide