关键词: 近似并行子空间优化, 功能测度法, 角度更新, 多学科设计优化, 可靠性分析

Abstract: To improve the computational efficiency of Reliability-based Multidisciplinary Design Optimization (RBMDO),a sequential RBMDO method based on an Approximate Concurrent Subspace Optimization strategy (ACSSO) and an Angle updating-based Performance Measure Approach (APMA) was proposed,which integrated decoupled idea and approximate techniques.This method decomposed RBMDO into a series of Deterministic MDO (DMDO) and multidisciplinary reliability analysis,which avoided the repeatedly calculations of reliability analysis model.DMDO could be implemented by ACSSO,in which the actual sensitivity was replaced by approximate sensitivity information and only calculated the sensitive index to model the global sensitivity analysis.As a result,all the non-sensitivity terms of discipline and system level were avoided.In addition,the efficiency of multidisciplinary reliability analysis was improved by APMA,in which the function value was replaced by angle updating of limit station function searching to reduce the calculations of coupling state variables and multidisciplinary analysis.The efficiency and engineering applications of proposed method was verified by an example of vehicle side impact design.

Key words: approximate concurrent subspace optimization, performance measure approach, angle updating, multi-disciplinary design optimization, reliability analysis