A representative product configuration ranking approach considering requirement interactions and inconsistent group preferences
针对产品配置中设计需求交互和多个利益相关者偏好不一致的问题，提出一种基于Choquet积分和稳健排序的配置方案排序方法，通过调整不一致偏好和构建代表性排序模型，选出兼顾个体与群体偏好的稳健方案。
Configuration solving is essential in product configuration because it supports decisions on selecting final configuration alternatives to satisfy stakeholders' preferences. However, the inherent interaction of product Design Requirements (DRs) and multiple stakeholders involved lead to three limitations in developing an appropriate configuration solving model: (i) the existing models have limitations in balancing the tractability of the model and the diversity of interactions; (ii) it is impractical to assume that different stakeholders have consistent preferences for the alternatives, and directly removing inconsistent preferences can lead to the loss of preference information; (iii) the representativeness of preference functions built from incomplete preference information is doubtful. To address these limitations, a robust product configuration ranking approach is proposed. First, the Choquet integral with the k-interactive fuzzy measure is applied to evaluate alternatives with DRs’ interactions, improving the tractability by reducing the number of parameters and related constraints. Second, an interactive method of identifying a minimum set of adjustable inconsistent preferences is proposed to adjust them among different stakeholders, avoiding the preferences loss and ensuring the trade-off among stakeholders. Third, a representative ranking model based on the k-interactive Choquet integral and nonadditive robust ordinal regression is proposed to select the robust alternative concerning individual and group stakeholders' preferences. To validate its feasibility and effectiveness, the proposed approach is applied to an example of configuration solving for a robotic manufacturing system.