A multi-agent method for adaptive real-time train scheduling with conflict limitations

Abstract

The paper proposes a multi-agent method to creating an intelligent adaptive system of train real time scheduling with conflict limitations. The architecture of the multi-agent system consisting of two base planning subsystems is described. Subsystem interaction protocols and protocols of agent interaction within each subsystem are presented. The example of schedule planning in various situations is presented in details. Realizable characteristics of the developed multiagent system are presented. High quality of schedule planning and system performance is shown.