Linguistic information-based granular computing based on a tournament selection operator-guided PSO for supporting multi-attribute group decision-making with distributed linguistic preference relations

Abstract

This study investigates linguistic information-based granular computing for supporting multi-attribute group decision-making where relative importance of experts/attributes, and outcomes of pairwise comparisons over alternatives are recorded in the form of distributed linguistic preference relations. First, we propose two linguistic information granulation models based on the measures of consistency and consensus to realize the operability of two types of linguistic information, respectively. The granulation of the weight linguistic information is closely linked with that of the alternative preference linguistic information in a way that two sets of control factors are introduced. Second, an estimation method for expert/attribute weights is developed based on the optimal control factors and solutions to the aforesaid linguistic information granulation models. The proposed linguistic information granulation models are in fact constrained non-linear optimization problems which may be neither effectively nor efficiently solved by general optimization methods. Third, we therefore propose a novel solution approach named tournament selection operator-guided particle swarm optimization (TSOG-PSO). Compared with the standard PSO and its variants, the TSOG-PSO enjoys more feasibility and lower complexity. A new energy vehicle evaluation problem is analyzed using the proposals to demonstrate their applicability. Some discussions and comparative studies are conducted to illustrate the effectiveness of the proposals.