Role of Emergent Computing Techniques in Multidisciplinary Rotor Blade Design
- Year
- 이전
- Author
- P. Hajela, J. Lee
- Journal
- Computer and Systems Sciences NATO ASI Series F
- Vol.
- Vol.149, p.162-187
- Issue Date
- 1996.01
Design synthesis in multidisciplinary systems is typically characterized by a large number of design variables and constraints; additionally, there are complex interactions between the participating disciplines which must be both identified and then suitably represented in the design process. In a number of applications, the design space itself may be multimodal, thereby introducing the need for a global search strategy; the latter may contribute to an excessive demand on computational resource requirements. This paper discusses some of the above problems in the context of a multidisciplinary design of a helicopter blade, where the underlying analysis is inherently nonlinear, computationally demanding, and the design space is nonconvex. The use of newly emergent computing techniques such as genetic algorithms and neural networks is discussed in the context of this problem. The role of neural networks as a function approximation strategy is discussed; the multilayer backpropagation network (BP network) and a variant of the counterpropagation network (CP network) are examined. The paper also shows how a trained BP network can be used to identify couplings in a multidisciplinary system. Such an identification provides a rational approach for determining the topology of probelm decomposition, and a sequence of smaller, more tractable subproblems can be generated.