CAREER: Development of Physical Programming for Robust Computational Design

点击翻译按钮获取中文摘要

This grant provides funding for the development of a new computational optimization method for engineering design called `physical programming`. This method is intended to make it possible to optimize physical systems in a computational environment that does not require the use of numerical weights on system attributes. The correct numerical values of these weights, which are typically difficult to identify, reflect the degree of conflict between the different design objectives. Rather than using weights, physical programming allows the designer to express his/her preferences by specifying several ranges of differing degrees of desirability for each design objective. This new approach is intended to bring simplicity, and reduced computational burden to the optimization process. The theoretical foundation of physical programming and related computational algorithms will be developed, together with optimization visualization methods that exploit the physical programming formalism. This grant also calls for the development of a practice-oriented course in design optimization for seniors and first-year graduate students. The objective of this course is to equip students to effectively use optimization in industrial settings, using physical programming. If successful, the outcome of this research will result in the development of physical programming to the point of commercial and industrial applicability. In recent years, industry has significantly increased its use of optimization to improve product designs. Presently, the effective and reliable use of optimization is not within the reach of most industry engineers on a day-to-day basis. Two main reasons are: (1) the required level of expertize in optimization, and (2) the difficulty of expressing the designer's preferences using weight-based methods. The proposed research will address the above-stated difficulties; and will also provide a new and more flexible framework to address linear and nonlinear programming problems.