Multi-agent Systems Modeling Based On Modelica And The Rotating Consensus Problem

Multi-Agent Systems(MASs), as an important research direction of artificial intelligence, has become a significant subject in the 21 st century. Meanwhile, the distributed cooperation of MASs has attracted a large number of attentions recently due to its wide applications. Consensus problem is not only the typical issue of agents’ cooperation control but also the fundamental problem. Roughly speaking, consensus problem can be described as a solution way to reach a certain common state by some designed control protocols of the agents.This paper focuses on the study of agents. The contents include the MASs modeling by Modelica langusge, putting forward a new method to analyze rotating consensus problem, and setting up a result display platform based on Visual Studio 2010. The main works and obtained results are discribed as follows:First, a MASs model library using Modelica language is set up. The model library includes agents’ mathematical model library, control algorithm library, sensor, interaction, communication and so on. Single agent and MASs can be modeled by those components in the sub- libraries mentioned above, and curve simulation results of models can be obtained. The biggest advantage of Modelica modeling is its desperation of controller and control object, which improves the reusability of component significantly and make the modeling of complicated MASs simpler.Then, the rotating conse nsus probelm of linear MASs is dealt with, which has wide applications, such as the cooperative motion that vehicles drive through a circle road, the flight of satellite aroud the earth. This paper puts forward a new method to solve the rotating consensus problem of MASs. First, the rotating consensus problem of the original system is transformed into a state consensus problem of a new system by a model transformation. Then, a proper linear transformation is proposed to transform the state consensus problem of new system into a partial stability problem of corresponding auxiliary system. Meanwhile, a necessary and sufficient condition is derived and the consensus function is given. What’s more, the gain matrix of the protocol is designed by solving a bilinear matrix inequality via the path-following method. At last, numerical examples under directed and undirected communication topologies are given to illustrate the effectiveness of the theoretical results.The last part sets up a MASs dynamic dispaly platform using Visual Studio 2010 to make the simulation results vivid and direct. The dispaly platform can read data file from the Modelica modeling platform and the data drive the agent to move.