Методы применения многоагентной системы для оценки состояния слабоформализуемой системы
Аннотация
В статье предлагается результаты исследования различных многоагентных систем на примере ряда моделей и многоагентных реализаций оптимизационных алгоритмов. Выделены общие методы построения и вопросы, касающихся их поведения, критерии качества работы систем. Определены закономерности, взаимосвязей между свойствами и параметрами, используемыми при задании многоагентной интеллектуальной системы. Разработаны подходы обработки сложноструктурированной информации. Разработаны алгоритмы построения многоагентной интеллектуальной системы.
In the article the results of a study of various multi-agent systems using the example of a number of models and multi-agent implementations of optimization algorithms are presented. The essence of multi-agent technology is a fundamentally new method of solving problems. In contrast to the classical method, when some well-defined (deterministic) algorithm is searched, which allows finding the best solution to a problem, in multi-agent technologies, the solution is automatically obtained as a result of the interaction of many independent targeted software modules — so-called software agents. The main idea of the proposed multi-agent system is based on the division of the entire system into consistently acting, autonomous intelligent agents. These agents compete and cooperate with each other in order to find a complete solution to a global problem and to carry out a synthesis of individual solutions to a common problem into a final solution. The proposed multi-agent distributed intelligent system consists of knowledgebased agents. Input information for all five agents is represented by fuzzy variables. Using fuzzy inference rules, each agent generates its own output solutions. General construction methods and issues related to their behavior, criteria for the quality of system performance are highlighted. The regularities, interrelations between the properties and parameters used when specifying a multi-agent intelligent system are defined. Developed approaches for processing complexly structured information. Developed algorithms for building multi-agent intellectual system.
In the article the results of a study of various multi-agent systems using the example of a number of models and multi-agent implementations of optimization algorithms are presented. The essence of multi-agent technology is a fundamentally new method of solving problems. In contrast to the classical method, when some well-defined (deterministic) algorithm is searched, which allows finding the best solution to a problem, in multi-agent technologies, the solution is automatically obtained as a result of the interaction of many independent targeted software modules — so-called software agents. The main idea of the proposed multi-agent system is based on the division of the entire system into consistently acting, autonomous intelligent agents. These agents compete and cooperate with each other in order to find a complete solution to a global problem and to carry out a synthesis of individual solutions to a common problem into a final solution. The proposed multi-agent distributed intelligent system consists of knowledgebased agents. Input information for all five agents is represented by fuzzy variables. Using fuzzy inference rules, each agent generates its own output solutions. General construction methods and issues related to their behavior, criteria for the quality of system performance are highlighted. The regularities, interrelations between the properties and parameters used when specifying a multi-agent intelligent system are defined. Developed approaches for processing complexly structured information. Developed algorithms for building multi-agent intellectual system.