Voltage control method considering virtual power plant in distribution network based on deep learning long and short-term memory networks

With the grid connection of virtual power plant in distribution network such as large-scale distributed power sources, energy storage devices and active loads, the traditional distribution network is transformed into a new type of active distribution network, which affects many aspects such as power quality and safe operation of the power grid, and also brings great challenges to the planning, optimization, control and management of the distribution network. The existing analytical algorithms and intelligent algorithms for solving the optimization and control problems of active distribution networks reveal many shortcomings such as complex mathematical modeling, difficulty in satisfying the solution accuracy and fast solution speed in the face of multiple optimization objectives and complex network structures, while deep learning algorithms have unique advantages in solving complex nonlinear problems. Therefore, this paper takes active distribution network voltage control as the research objective, considers the power output characteristics of distributed power sources and energy storage devices and the grid operation constraints, designs a multi-time scale voltage control strategy, including two time phases before and during the day, schedules the reactive power output of distributed power sources and the charging and discharging of energy storage systems, and uses a deep learning long and short-term memory network as the solution algorithm. Using the improved 33-node distribution system as an example, the results verify the feasibility of the long and short-term memory network for optimization calculation, which can effectively control the node voltage, reduce the active network loss, and ensure the accuracy of the optimization control without the complex mathematical model, and at the same time can realize fast dispatching.