Traditional robust design commonly uses a weighted sum to reconcile nominal performance with variability. This paper addresses the shortcomings of this use of the weighted sum in robust design. A weighted sum can fail to obtain the desired design for given trade-off between mean and variability of performance; it may not be able to reach all the Pareto points, and even when it can reach all Pareto points, there is no guarantee that it selects the desired one. To overcome this inadequacy, a preference aggregation method is incorporated into the classical robust design method. This preference aggregation approach is able to select any Pareto point. In addition, prior work on preference aggregation includes a rigorous method for determining the decision trade-off parameters and thus the desired Pareto solution, without the need to recover the entire Pareto frontier. The proposed method is also compared with compromise programming (CP) for robust design. Despite many superficial similarities, the two methods are fundamentally different. The compromise programming approach relies upon an efficient algorithm to generate an entire Pareto frontier of robust designs, leaving the user to select a desired solution among them. A case study of the use of compromise programming, weighted sums, and preference aggregation methods for generating a Pareto frontier indicates that preference aggregation is as computationally efficient as the other methods but gives better-spaced results. In addition, the preference aggregation method allows users to determine what are good trade-off parameters, and to use these parameters to help guide a design or to explore potential Pareto designs without the computational burden of recovering the entire Pareto frontier. The indifference point calculation presented in detail in prior publications helps avoid arbitrary decisions on specification of importance weights and trade-off parameters. In the procedure to solve the solution for the problem of reuse of bumper, all three Kriging models are involved in the optimization run. Such efficient and appropriately accurate approximation models are often required in the design reuse problem, especially when the size of the problem increases.


    Zugriff

    Zugriff über TIB

    Verfügbarkeit in meiner Bibliothek prüfen

    Bestellung bei Subito €


    Exportieren, teilen und zitieren



    Titel :

    Improving robust design with preference aggregation methods


    Beteiligte:
    Dai, Z. (Autor:in) / Scott, M.J. (Autor:in) / Mourelatos, Z.P. (Autor:in)


    Erscheinungsdatum :

    2004


    Format / Umfang :

    15 Seiten, 22 Bilder, 3 Tabellen, 24 Quellen




    Medientyp :

    Aufsatz (Konferenz)


    Format :

    Print


    Sprache :

    Englisch




    Improving Robust Design with Preference Aggregation Methods

    Dai, Z. / Scott, M. J. / Mourelatos, Z. P. et al. | British Library Conference Proceedings | 2004


    Improving Robust Design with Preference Aggregation Methods

    Dai, Z. / Scott, M. J. / Mourelatos, Z. P. et al. | British Library Conference Proceedings | 2004


    Improving Robust Design with Preference Aggregation Methods

    Scott, Michael J. / Dai, Zhihuang / Mourelatos, Zissimos P. | SAE Technical Papers | 2004


    Improving robust design with preference aggregation methods

    Dai,Z. / Scott,M.J. / Mourelatos,Z.P. et al. | Kraftfahrwesen | 2004