Abstract

A qualitative reasoning approach is presented for decision support. This approach is initiated with the generation and normalization of a qualitative model including relationship matrix, importance weightings, control limits for decision variables, and specification limits for performance attributes. An extensive Simplex method is then utilized for mapping the global feasibility of the decision problem. Finally, the approach interprets and simplifies the results from the multi-dimensional feasibility space back to the qualitative model. The paper also presents application of the approach to process tuning of multilayer sheet extrusion and concept development of a 2nd generation reusable launch vehicle to orbit.

The results indicate that the developed approach is a viable alternative to currently practiced decision analysis methods. Specifically, the approach uses: 1) tightened specification limits rather than a weighted multi-attribute objective relationship, 2) determines global system feasibility from active constraints rather than local parametric optimization, and 3) establishes correlations based on dynamic Pareto-optimal frontiers rather than manual estimations. While the approach is potentially useful, further research is required to extend the methods to non-linear, stochastic, and uncharacterized engineering systems.

This content is only available via PDF.
You do not currently have access to this content.