Integrated Model for Constructing Evolving Multi-Agent Subsystems
Гладков Леонид Анатольевич, Гладкова Надежда Викторовна, Надежда Алексеевна Дмитриенко · 2019 International Russian Automation Conference (RusAutoCon) · 2019
The paper provides evolutionary design methods for autonomous agents and multi-agent systems (MAS). Hybrid and fuzzy models development is based on design methods. Evolutionary principles and models, especially evolutionary design techniques, are used in the implementation of multi-agent systems. The agent and MAS design principle was developed. According to the principle, a design process includes basic self-organization components like interaction, crossing, adaptation to environment, etc. A general technique is developed: descendant agents are formed as a result of parent agent interaction. Different types of crossing-over are developed and described, also the idea of forming agencies (families) as evolving MAS simples was created. According to fuzzy coding principles, hybrid fuzzy and evolutionary models of forming agents and agencies were developed. Based on the fuzzy logic approaches such as triangle norms and conorms, fuzzy genetic operators development methods were provided. A program support system for agents and MAS evolutionary design was developed and implemented.