Research On Optimal Scheduling Of Virtual Power Plants Considering Carbon Trading Under The Dual Carbon Target

In the background of the global temperature continues to rise,environmental pollution intensified,China’s power industry has accelerated the pace of clean and lowcarbon development,renewable energy generation installed capacity continues to grow rapidly.However,the renewable energy generation represented by wind power and photovoltaic has volatility and uncertainty.The high percentage of renewable energy connected to the grid is a major challenge to the safe and stable operation of the power system.Electric vehicles,energy storage systems,adjustable loads and other flexible resources have great potential and can play an extremely important role in regulating the instability of renewable energy output and ensuring the reliability and continuity of power supply to the grid.As a special power plant,the virtual power plant(VPP)can effectively integrate distributed wind power,photovoltaic,energy storage,gas units and other flexible resources to achieve stable and reliable power output through internal coordination and scheduling.In addition,some of the generating units in the VPP will still have some carbon emissions.The initiative of China’s carbon trading mechanism offers the possibility to further reduce carbon emissions from the VPP.Therefore,in the context of the dual carbon target,it is important to study the optimal dispatching method of the VPP participating in carbon trading to realize the near-zero carbon emission operation of the VPP and the safe and stable operation of power systems.This paper studied the optimal scheduling of the VPP participating in carbon trading.First.on the basis of reviewing a large amount of literature,the research dynamics of domestic and foreign scholars on the VPP optimal scheduling model,optimal scheduling solution algorithm,carbon trading,and power forecasting were sorted out.Second,the underlying theories of carbon trading mechanism,data space and deep reinforcement learning were introduced in detail to lay the foundation for the later research.Third,the basic concept and basic structure of the VPP were defined,the current situation of the VPP application at home and abroad was sorted out,and the scheduling process of the VPP participation in the power market was analyzed on this basis.This was followed by a description of how the VPP can participate in carbon trading,giving a basic framework for the VPP to participate in both the electricity market and the carbon market.Fourth,in response to the complexity of the internal units of the VPP and the characteristics of massive and multi-source heterogeneous data,the data space was constructed to improve the data utilization and management efficiency of the VPP.In addition,it was clear from the combing of the operating mode that accurate forecasting of wind and photovoltaic power generation was an important prerequisite for optimal scheduling of the VPP.Therefore,this paper constructed a combined power prediction model based on convolutional neural network-long and short-term memory neural network(CNN-LSTM).And the data were extracted from the VPP data space for wind power and photovoltaic power forecasting to verify the effectiveness of the constructed combined forecasting model.Fifth,based on the operation process of the VPP participating in the power market and carbon market,a two-stage of day-ahead-intra-day optimal scheduling model for the VPP participating in carbon trading under the dual carbon target was constructed.The model aimed at profit maximization and carbon emission minimization,considered energy gain,carbon gain,operation cost and deviation penalty cost,and was solved by the DQN algorithm in the deep reinforcement learning algorithm.Sixth,four different scenarios were set up for simulation analysis to verify the superiority of the optimal scheduling model proposed in this paper.Among them.the predicted values of wind power and photovoltaic power were taken from the predictions of the combined CNN-LSTM model.And the applicability of DQN algorithm in solving the optimal scheduling model of the VPP was verified by comparing different algorithms.In this paper,the wind and photovoltaic power prediction models based on the VPP data space and the combined CNN-LSTM algorithm achieved high prediction accuracy,and the two-stage optimal scheduling model for the VPP considering carbon trading under dual carbon targets improved the revenue while limiting carbon emissions.These can provide model methods for the accurate prediction of distributed power generation power of the VPP and the decision of optimal scheduling scheme,which help the VPP stabilize power output,improve new energy consumption and reduce carbon emissions.These provide a reference for the operation and management of the VPP,thus helping to promote the clean and low-carbon development and safe and stable operation of power systems.