Large-scale group consensus hybrid strategies with three-dimensional clustering optimisation based on normal cloud models

Abstract

Large-scale group decision-making (LSGDM) is characterised by a large number of experts and a complex consensus reaching process. Clustering is used to divide the large group into a number of manageable subgroups; however, the simultaneous presence of all subgroup members at the negotiation process is rare. Thus, the selection of subgroup representatives for a smooth negotiation is necessary. Few LSGDM consensus recommendation optimisation models truly consider the problems of subgroup representative selection in their strategy to reach a consensus. This article proposes LSGDM consensus hybrid strategy framework with three-dimension clustering optimisation based on normal cloud models (NCMs) whose aims are threefold: (1) the use of NCMs to represent the imprecision of linguistic preferences provided in real complex decision scenarios with a large number of experts; (2) to establish a clustering optimisation method to choose subgroup representatives using three sensible criteria: preference similarity level within the subgroup, preference precision level, and preference consistency level; and (3) to establish two consensus recommendation optimisation strategies for individual negotiation-guided and moderator-guided consensus reaching, respectively. The feasibility and applicability of the proposed method are illustrated via a power curtailment policy assessment example, then some sensitive and comparative analyses are conducted to explicit the effectiveness and advantages of the proposed consensus hybrid strategies.