Reliability Based Robust Design Optimization Using Performance Measure Approach

• Main Authors: Jai Pon Yang, 楊家鵬
• Other Authors: 史建中
• Format: Others
• Language: zh-TW
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/73342306869790858359

碩士 === 淡江大學 === 機械與機電工程學系碩士班 === 98 === In general, a reliability-based design optimization (RBDO) problem contains the system reliability consideration of optimizing design goal subject to design constraints. Due to the uncertainty of random variables and/or random parameters, the final design may have a low possibility of satisfying constraints. In other word, once the design closes to constrained boundary (means active constraint), it may have a high possibility in infeasible region. This thesis applies a variation sphere transformation that moves the feasibility boundary to promote the feasibility of active constraints. The complete design process is called the RBDO considering feasibility robustness (RBDO-FR). Due to the uncertainty of random variables and/or random parameters, the final design may also have a wide distribution of design goal performance. In other word, the larger deviation of performance is not a result of robust performance. This thesis applies a variation sphere transformation, two-level factorial experiment and Takuchi three-point approximation to estimate the variation. We minimize the performance variation so that the complete design process is called the RBDO considering performance robustness (RBDO-PR). Base on the consideration of RBDO-FR and RBDO-PR, it is obviously to take into account a RBDO problem simultaneously contain both feasibility robustness and performance robustness. Thus, a RBDO with total robustness (RBDO-TR) is the main development in this thesis. The whole integrated design process of RBDO-TR is presented with some illustrative examples and engineering design. In the development, the performance measure approach (PMA) is applied for the structural reliability analysis.