Abstract:

Multi-attribute group decision-making （MAGDM） is widely used in many social fields. In practical MAGDM， the heterogeneity between decision makers （DMs） or attributes causes it to be difficult to accurately describe all attribute information in the form of a single preference. Meanwhile， there are often complex interactions between attributes or alternatives， which， if ignored， can easily lead to inaccurate attribute weights and preference information fusion， reducing decision quality. Moreover， due to the psychological influence of the decision environment and subjective cognitive limitations on DMs， DMs are often in a state of incomplete rational psychology during the decision-making process， and this can have a significant impact on the decision results. Therefore， it is necessary and important to consider the heterogeneous information， attribute and alternative interaction， and DMs' incomplete rational psychology in MAGDM. However， few MAGDM methods consider all three aspects simultaneously.To make this gap， a new MAGDM method in a fuzzy heterogeneous environment considering dual interactions and DMs' incomplete rational psychology is proposed. First， a bidirectional projection measurement of fuzzy heterogeneous information is defined， and the homogenous method of three-dimensional D-U space based on bidirectional projection is established. Second， a variance maximization optimization model is constructed to determine the local weights of each interaction attribute， and then the global interaction weights of interaction attributes and attribute sets are determined based on the generalized Banzhaf values of the λ fuzzy measure. Then， to describe and measure the alternative interactions， a alternative complex interaction network of integrated perceptual dominance and a global interaction integrated perceptual dominance model are developed. Following this， the classical TODIM model is extended， and an attribute and alternative dual interaction TODIM model considering the imperfect rational psychology of DMs is proposed to determine the alternative ranking. Finally， the validity and feasibility of the proposed method are proved by a site selection case study of hot dry rock development， and its superiority and scientificity are verified by comparative analysis. The results show that simultaneous consideration of heterogeneous information， attribute and alternative interactions and DMs' incomplete rational psychology will substantially improve the accuracy and robustness of MAGDM methods.

Key words: multi-attribute group decision making, fuzzy heterogeneous information, attribute and alternative double interaction, incomplete rational psychology, TODIM model