Research And Application Of Game Theory On Power Market Bidding Behavior

A graded game model based on multi-agent platform RePast is proposed to meet the requirement that power generation enterprises need to predict competitors' information accurately in an incomplete information and monopolistic power generation market.The model predicts competitors' information with experiential values at the beginning when information is less;with the increase of game times,the model will continually modify its predict functions according to the information acquired by far so as to upgrade its prediction accuracy,which demonstrates the capability that power generation enterprises will make full use of the information accumulated during the game process for continual modification of their game behavior rules.The simulation of a simplified power market, assumed to be consisted of power generation enterprises with same kind of production cost functions,is carried out on RePast platform.The simulation results demonstrate that the graded game model can predict the competitors' information more accurately than others.