We have developed a data visualization interface that facilitates a design by shopping paradigm, allowing a decision-maker to form a preference by viewing a rich set of good designs and use this preference to choose an optimal design. Design automation has allowed us to implement this paradigm, since a large number of designs can be synthesized in a short period of time. The interface allows users to visualize complex design spaces by using multi-dimensional visualization techniques that include customizable glyph plots, parallel coordinates, linked views, brushing, and histograms. As is common with data mining tools, the user can specify upper and lower bounds on the design space variables, assign variables to glyph axes and parallel coordinate plots, and dynamically brush variables. Additionally, preference shading for visualizing a user’s preference structure and algorithms for visualizing the Pareto frontier have been incorporated into the interface to help shape a decision-maker’s preference. Use of the interface is demonstrated using a satellite design example by highlighting different preference structures and resulting Pareto frontiers. The capabilities of the design by shopping interface were driven by real industrial customer needs, and the interface was demonstrated at a spacecraft design conducted by a team at Lockheed Martin, consisting of Mars spacecraft design experts.

Recent advancements in computing power and speed provide opportunities to revolutionize trade space exploration, particularly for the design of complex systems such as automobiles, aircraft, and spacecraft. In this paper, we introduce three Visual Steering Commands to support trade space exploration and demonstrate their use within a powerful data visualization tool that allows designers to explore multidimensional trade spaces using glyph, 1-D and 2-D histogram, 2-D scatter, scatter matrix, and parallel coordinate plots; linked views; brushing; preference shading and Pareto frontier display. In particular, we define three user-guided samplers that enable designers to explore (1) the entire design space, (2) near a point of interest, or (3) within a region of high preference. We illustrate these three samplers with a vehicle configuration model that evaluates the technical feasibility of new vehicle concepts. Future research is also discussed.

Recent advancements in computing power and speed provide opportunities to revolutionize trade space exploration, particularly for the design of complex systems such as automobiles, aircraft, and spacecraft. In this paper, we introduce three visual steering commands to support trade space exploration and demonstrate their use within a powerful data visualization tool that allows designers to explore multidimensional trade spaces using glyph, 1D and 2D histograms, 2D scatter, scatter matrix, and parallel coordinate plots, linked views, brushing, preference shading, and Pareto frontier display. In particular, we define three user-guided samplers that enable designers to explore (1) the entire design space, (2) near a point of interest, or (3) within a region of high preference. We illustrate these three samplers with a vehicle configuration model that evaluates the technical feasibility of new vehicle concepts. Future research is also discussed.

The authors present a Generative Adversarial Network (GAN) model that learns how to generate 3D models in their native format so that they can either be evaluated using complex simulation environments, or realized using methods such as additive manufacturing. Once initially trained, the GAN can create additional training data itself by generating new designs, evaluating them in a physics-based virtual environment, and adding the high performing ones to the training set. A case study involving a GAN model that is initially trained on 4045 3D aircraft models is used for demonstration, where the training data set updated with GAN-generated and evaluated designs, results in enhanced model generation, both in the geometric feasibility and performance of the designs. Z-tests on the performance scores of the generated aircraft models indicate a statistically significant improvement in the functionality of the generated models after three iterations of the training-evaluation process. In the case study, a number of techniques are explored to structure the generate-evaluate process in order to balance the need to generate feasible designs with the need for innovative designs.

No abstract