Multi-objective Robust Optimization Consensus Model for Large-Scale Group Decision Making
Abstract
In large-scale group decision making (LSGDM), traditional clustering methods typically rely on a single distance metric, ignoring both intra-cluster compactness and inter-cluster separation, and often fail to adaptively determine the optimal number of clusters. In addition, most existing consensus models, while controlling adjustment costs, do not adequately consider the improvement of the global consensus level and ignore the uncertainty inherent in unit adjustment costs. To address these issues, this paper first develops a multi-objective clustering algorithm (MOCA) based on NSGA-II, in which intra-cluster compactness and inter-cluster separation are jointly considered as two optimization objectives. By searching over different numbers of clusters and incorporating a minimum cluster-size mechanism, MOCA identifies a suitable expert partition while maintaining both subgroup cohesion and differentiation. Subsequently, a multi-objective robust optimization model is established to adjust expert opinions, with the objectives of minimizing the total adjustment cost while simultaneously maximizing the global consensus level, where robust counterparts of the constraints are incorporated to capture uncertainties in unit adjustment costs. Finally, a financial investment decision-making example is provided to demonstrate the practicality and effectiveness of the proposed model and the sensitivity analysis is also conducted.