Оптимизация стратегий поведения в имитационной модели многоагентной социально-экономической системы

Экономика и математические методы. 2023. Т. 59. № 3. С. 117–131.

Akopov A. S., Beklaryan A.

This article presents a new simulation model of a multi-agent socio-economic system (MA-SES), in which individual, including interproduct interactions are implemented. Within the MA-SES, the models of agent-producers and agents-consumers behaviour with their objective functions are studied. In particular, producers follow their own production strategies by choosing the moments for introducing new fixed assets and labour resources. Consumers participate in barter and monetary deals by interacting with other agents. The states of readiness of agents-producers to introduce new fixed assets and labour resources, as well as the states of readiness of agents-consumers to complete barter and monetary deals are set up for each time moment using lognormal distributions with given characteristics, which are the control parameters of the system. Important bi-objective optimisation problems are formulated for ensembles of agents-producers and agents-consumers. A new parallel hybrid genetic algorithm (MORCGA-MOPSO) was developed, in particular, providing the possibility of seeking the Pareto-optimal solutions for maximising the average (over an ensemble of agents) profit and the total number of agent-producers’ buyers, as well as maximising the average utility and monetary savings of agents-consumers. The parameters of log-normal distributions that determine the states of interacting agents that make individual decisions are computed. The features of the behavioural strategies of producers and consumers are determined, which make possible to achieve an improvement in the values of the objective functions through controlling the dynamics of the input of production resources and choosing the preferred types of interproduct interactions, in particular, barter, monetary etc.

Keywords: многокритериальная оптимизация, multicriteria optimization, genetic algorithms, генетические алгоритмы, Particle swarm optimization, метод роя частиц, multi-agent socio-economic systems, simulation of economic processes, interproduct interactions, многоагентная социально-экономическая система, межпродуктовые взаимодействия, имитационное моделирование экономических процессов