Research on optimization strategies for the operation of multiple transformer districts con-sidering the uncertainty of distributed renewable energy output and carbon emission costs
With the market-oriented reform of power grid companies,the power market will gradually attract the investment of various social capital.The transformer districts(TDs)subordinated to the distribution network and the distribution network itself provided a platform for the multi-agent competition,forming a competitive game pattern.At the same time,the high proportion of DRE access improves the cleanliness of the distribution network,but the uncertainty of DREs'output also leads to the further increase of the distribution network dispatching operation risk.To mitigate the uncertainty,the distributed renewable energy,distributed thermal power generation,energy storage and flexible load within the same TD is treated as a whole and regulated by the distribution grid operator with the objectives of safety and economy.Firstly,a leader-follower game model consists of the distribution grid operator and multiple transformer districts is established to coordinate the interests between the distribution grid operator and its subordinate TDs.Conditional value-at-risk theory is used to quantify the uncertainty risk caused by renewable energy represented by wind and solar power.Next,the profit of each TD in the carbon market is incorporated into the optimization scheduling model to further consider the carbon emission costs of distributed thermal power generation achieving flexible complementary regulation between distributed renewable energy and thermal power.The BP neural network is used to fit the model,simplifying the leader-follower game model into a single-level model,which is then solved using a particle swarm algorithm.Finally,the variations in dis-tributed power generation within each TD under different renewable energy output risks and carbon prices are discussed to further validate the effectiveness of the model.
uncertaintycarbon emission costsconditional value-at-riskBP neural networkstackelberg game
万方数据
维普
