Multidisciplinary Design Optimization Theory

TL;DR: In this article, the authors introduce the iconic approach and other key technologies for multidisciplinary design optimization, including mathematical modeling, design-oriented analysis, approximate technology, optimization processes, system sensitivity analysis, and artificial interfaces.

Abstract: Sobieski, the founder of multidisciplinary design optimization, believes that there are six key technologies for multidisciplinary design optimization: mathematical modeling, design-oriented analysis, approximate technology, optimization processes, system sensitivity analysis, and artificial interfaces (Sobieszczanski-Sobieski and Haftka 1996), the most important of which is mathematical modeling. The method of dealing with the coupling relationship between systems in mathematical modeling is called multidisciplinary design optimization method. The method to deal with the coupling relationship between systems in mathematical modeling is called multidisciplinary design optimization method, which is the research focus of multidisciplinary design optimization theory. According to whether the optimization is carried out at single-level or at multiple levels, the multi-disciplinary design optimization methods can be divided into single-level optimization algorithm and multi-level optimization algorithm. The single-level multidisciplinary design optimization method does not decompose the original system model, but only carries out optimization at the top level of the system, and achieves subsystem balance through iteration among subsystems or sub-disciplines. Common single-level multidisciplinary design optimization methods include Multidisciplinary Feasible Method (MDF), Individual Disciplinary Feasible Method (IDF), and Successive Approximate Optimization (SAO). The multilevel optimization algorithm is optimized at the system level and in the sub-discipline, and the sub-discipline optimization is conducted around the system optimization. This algorithm not only conforms to people’s thinking mode of decomposing a problem into several sub-problems, but also can execute in parallel. Common multi-level optimization algorithms include Collaborative Optimization (CO), Concurrent Subspace Optimization (CSSO) and Bi-Level Integrated Synthesis (BLISS). This chapter introduces the iconic approach and other key technologies for multidisciplinary design optimization.