Reactive Power Optimization Based on Multi- agent System with Markov Game Theory

This paper presents a distributed and reactive power optimization system based on Muti-Agent System framework. The various component and functions of monomer Agent are displayed in the optimization system. The collaboration and coordination between the Agent are achieved by information sharing through the communication. The dynamic contract net protocol which doped by three mental state performance parameters contains Agent, trust nodes, familiarity and integration introduced into the collaborative approach of Muti-Agent. The optimization process of lower voltage and reactive power are described based on the Dynamic contract net protocol under the smart grid environment. The Muti-Agent learning algorithm is proposed according to the influence of the chosen Q value based on Markov game framework. The Reactive Agentbased design optimization process, management system architecture design, simulation protocol design are illustrated combined with Agent combines technical features and smart distribution network reactive power compensation behavioral characteristics. Keywords-markov game;muti-agent; power systems ; reactive power optimization