제한된 이산정보를 이용한 엔진 크래들의 신뢰성 기반 최적설계

Abstract

To achieve mass reduction and performance improvement, design optimization has been studied in automotive engineering. Recently, stochastic design optimization methods are used to calculate not only the performance but also a probability of the performance. Reliability-based design optimization (RBDO) is one of the stochastic design optimization methods. In RBDO, by considering stochastic information of input variables, the reliability of the performance can be estimated. In this paper, we adopt the RBDO to design an engine cradle with limited discrete information from tolerance of a panel of an engine cradle. With the stochastic information of the input variables, the reliabilities of the performance such as stiffness, life and permanent displacement are estimated by the reliability analysis using Akaike information criterion (AIC). AIC can find the best estimated distribution of the performance from limited discrete information and then estimate the reliability from the distribution. Finally, we can reduce the mass and satisfy the target reliabilities of the performances of the engine cradle.