Decomposition of complex systems into set of autonomous agents by fuzzy-genetic approach and its application in economic and business environments

Economic, ecology and business systems are often complex systems, and are almost always characterized by imprecision and uncertainty. It is known that in such cases distributed multi-agent intelligent system based on soft computing is the most effective approach for systems analysis, decision making, and control in such systems. The key problem in constructing such systems is the problem of granulation, i.e., decomposition of the monolith intelligence of the whole system into autonomous agents' intelligence. The work suggests a method for creation of optimal knowledge bases of coordinating and cooperating intelligent agents. The optimization includes determination of a rational number of autonomous agent and fuzzy rules, optimal scaling factors, shapes and centers of membership functions of fuzzy rules of agents' knowledge bases, and optimal inference engine by using genetic algorithms. Computer simulation of the multi-agent distributed system for marketing-mix decision support system and demand forecasting are provided.