Research On Bilateral Matching Model And Algorithm Based On Uncertain Preference Order

Bilateral matching widely exists in various fields of social life.It is the core problem of bilateral matching decision theory to fully explore the hidden information behind multi-form evaluation and express effectively and solve the matching model reasonably.In this thesis,for the case where the evaluation information is uncertain preference order,the regret theory and the principle of fairness are introduced into the matching model,and a self-adaptive learning sub-group parallel cooperative particle swarm algorithm is designed to solve.In solving the model of bilateral matching decision based on uncertain preference order,matching individuals often lack rational judgment,this thesis extract effective information in two aspects of competition and hesitation from the uncertain preference order as the true preference value of matching individuals.Taking account of the psychological factors of matching individuals,the regret theory is introduced to calculate the disappointment value and elation value of each matching pair,at the same time the psychological perception of the matching individual to all potential matching objects are also considered.Considering that in the actual matching process,there may be cases where an individual is difficult to accept the matching scheme.The matching pairs are given corresponding weights according to the principle of fairness and the degree of difference,reducing the weight of the matching pairs whose degree of difference is obvious,so that the matching scheme is relatively fair.In this thesis,a two-subgroup cooperative particle swarm algorithm is proposed to solve the bilateral matching decision model.In order to avoid falling into the local optimal solution too fast,the particle swarm is divided into ordinary subgroups and elite subgroups.Ordinary subgroups use the maximum fitness values of the last two generations as dynamic coefficients to guide particles to learn other excellent positions and individual worst positions.By introducing more learning sources,they avoid falling into local optimal solution.After the velocity and position of each particle are updated,the elite subgroups use the group's optimal position and the group's worst position to dynamically adjust the group's sub-optimal position for local search and learning,finally improving the accuracy of the algorithm.The simulation experiment results show that the introduction of regret theory improves the comprehensive perception utility value of the matching scheme,and the fairness principle reduces the degree of difference between the two parties of the matching objects.Compared with the basic particle swarm algorithm,the two-subgroup cooperative particle swarm algorithm can achieve a greater degree in jumping out of the local optimal solution,at the same time the comprehensive perception utility value of the matching scheme is significantly improved.