Scheduling Optimization of Master-slave Game in Microgrid Based on Bayesian Improved Deep Learning Algorithm under Dual Carbon Objective
Uncertainties in wind turbines,photovoltaic systems,and loads,along with the lack of coordination among multiple stakeholders,pose challenges to the optimization of microgrid dispatch.A multi-agent,multi-objective coordinated dispatch optimization model for microgrids is proposed based on BNN-DL source-load forecasting.Firstly,a deep neural network integrated with Bayesian methods is employed to map the nonlinear relationships among historical data,meteorological factors,and source-load,achieving accurate source-load forecasting.Secondly,considering the economics and energy utilization of microgrids,a two-stage model for day-ahead and real-time optimization is proposed.In the day-ahead stage,the model is optimized with the objective of minimizing the costs of energy storage operation,load dispatch,and electricity trading.In the real-time stage,to effectively coordinate microgrid energy interactions,a Stackelberg game model is established between microgrid generation and users.Additionally,non-cooperative games among generators and evolutionary games among users are employed to simulate electricity pricing and purchasing strategies,proving the existence of equilibrium solutions.The reverse mutation sparrow search algorithm is then used to solve the two-stage model.Finally,simulation analysis through case studies shows that the proposed method exhibits good adaptability,improving the economic benefits and energy utilization efficiency of microgrids.
Dual carbon objectivedeep learningBayesian algorithmmicrogrid scheduling optimizationmaster-slave game
万方数据
维普
