Interactive two response product and process design optimization under model parameter uncertainty

Date

2023-7-27

Author

Özateş Gürbüz, Melis.

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

119
views

95
downloads

Cite This

In this study, we address two response product and process design optimization problems. Finding meaningful solutions to these problems requires simultaneous consideration of the responses represented as empirical functions of relevant design parameters. Many of the existing approaches assume that the preferences of the decision maker (DM) are known prior to the solution procedure. This is a strong assumption that is hard to justify in practice. Other approaches incorporate DM’s preferences into the solution process in an interactive manner. Of these, only two approaches consider model parameter uncertainty in the empirical functions. We demonstrate the deficiencies of these two interactive approaches that consider model parameter uncertainty in detail with examples. To overcome the issues with the existing approaches, we develop two interactive optimization approaches considering DM’s preferences under model parameter uncertainty. We empower the DM to consider the area of the joint prediction region and the distances of the response means from their target values as performance measures. To facilitate the interaction with the DM, we produce visual aids on joint confidence and prediction regions of the responses at solutions under consideration. The main differences between our two approaches arise from the differences in the empirical model building. The first is for the case when a closed-form function for the joint prediction region is available, and the second is when it is not available. The latter complicates the process of navigating towards desirable solutions. We illustrate the developed approaches on a two response problem widely used in the literature.

Subject Keywords

Multi-response problems, Robust product and process design optimization, Multi-response surface optimization, Interactive multi-objective optimization approaches, Model parameter uncertainty

URI

https://hdl.handle.net/11511/104834

Collections

Graduate School of Natural and Applied Sciences, Thesis