Utilization of a dependency-tracking language to reduce computational time during multidisciplinary design optimization

Abstract

Reduction in computational time was desired and achieved in optimizing a multidisciplinary missile design system. The time reduction was realized using an object-oriented, dependency-tracking, demand-driven language called the adaptive modeling language (AML). The features of this language allowing for reduction in computational time are referred to as dependency-tracking and demand-driven computations. The dependency-tracking feature keeps track of the relationship amongst properties and objects within the hierarchy. This feature ensures that only necessary computations be carried out and it also ensures that computations that have previously been performed not be carried out again so long as the input to these computations have not changed. The timesaving features of this language make it an attractive choice when performing optimizations. A computational reduction in time of between 33 and 44% was achieved in the case when the language was used in conjunction with design of experiment and response surface models. The missile design system, interactive missile design and the optimization interface are coded in AML. The efficiency of the language was studied in conjunction with design of experiment, response surface analysis, and gradient-based optimization. The advancement of the missile design software by integrating optimization functionality is also discussed.