ASSESSMENT OF USER-GUIDED VISUAL STEERING COMMANDS DURING TRADE SPACE EXPLORATION

Open Access
- Author:
- Carlsen, Daniel Edward
- Graduate Program:
- Mechanical Engineering
- Degree:
- Master of Science
- Document Type:
- Master Thesis
- Date of Defense:
- None
- Committee Members:
- Timothy William Simpson, Thesis Advisor/Co-Advisor
- Keywords:
- ATSV
trade space exploration
visual steering
user-guided
interactive optimization
multi-objective genetic algorithm
visualization - Abstract:
- Trade space exploration is a promising decision-making paradigm that provides a visual and more intuitive means for formulating, adjusting, and ultimately solving design optimization problems. This is achieved by combining multi-dimensional data visualization techniques with visual steering commands to allow designers to “steer” the optimization process while searching for the best, or Pareto optimal, design(s). In this thesis, results from an empirical assessment of the performance of these visual steering commands are presented. This is done by performing a study that compares the performance of different combinations of these visual steering commands to automated samplers, including a multi-objective genetic algorithm, that are executed “blindly” on the same design problems with no human intervention. The resultant Pareto frontiers generated by the combinations of visual steering commands and automated samplers are compared to one another using the ε-performance metric to assess the extent to which they have identified the reference (or best known) Pareto frontier. Specifically, three test problems are examined: (1) a sandwich beam, (2) an aircraft wing, and (3) a vehicle configuration problem. The results of this study indicate that the visual steering commands – depending on the complexity of the test problem – can provide a 3x - 6x increase in the number of Pareto solutions that are obtained when the human is “in-the-loop” during the optimization process compared to an automated sampler. The improvements are even more dramatic in cases where automated samplers have a difficult time finding feasible solutions. In addition user-guided trials can provide from a 10x - 32x increase in the number of Pareto solutions obtained over random searching. As such, this study provides empirical evidence of the benefits of interactive visualization-based strategies to support engineering design optimization and decision-making.