Optimal scheduling model of a virtual power plant based on robust optimization with additional momentum to improve the predictive output of back propagation neural networks

In response to energy shortages, uneven distribution, and severe pollution, the global energy structure is rapidly changing. In the dispatching of power systems, the coordinated planning and flexible regulation of virtual power plants play a crucial role. This paper proposes a multi-objective model considering economic efficiency and carbon emissions to study the scheduling of virtual power plants and the proportion of new energy installed capacity. Firstly, the paper optimizes the power system load curve by implementing time-of-use pricing strategies, alleviating the additional pressure on installed capacity caused by demand differences during peak and off-peak periods. Secondly, an improved back propagation neural network method is employed to refine the robust interval, and by integrating feed-back historical data, the adaptive robust control theory is enhanced, thereby improving the system robustness and adaptability. Finally, through specific case analysis and scenario simulation, the paper finds that when the proportion of new energy in the system reaches 60%, it is possible to maximize economic efficiency and minimize carbon emissions while ensuring the stable operation of the virtual power plant.