Computational methods for optimum design of large complex systems (Q1820022)

Computational procedures for optimal design of large complex systems are described. Requirements of a good algorithm are discussed. A general design optimization model applicable to several classes of problems is defined. Several optimization algorithms are outlined and differences between them are highlighted. Modern algorithms generate and use approximate Hessian of the Lagrange function to calculate the search direction. They are quite reliable and become extremely efficient when a potential constraint strategy is incorporated into them. Based on recent experience with them, they are recommended for general engineering design applications. Several other computational aspects are also discussed, such as robust implementation of algorithms, use of knowledge base in providing consulting and diagnostic support to the designer, interactive use of optimization, and role of a database and database management system in design optimization.