Power distribution network reactive power optimization method and device based on data driving

Abstract

The invention discloses a power distribution network reactive power optimization method and device based on data driving, and the method comprises the steps: inputting historical data into a preset lower-layer neural network for training, and obtaining a lower-layer training model and unmeasurable key node voltage data; inputting a preset upper neural network for training according to the acquired voltage data of the measurable nodes and the non-measurable key nodes of the target regional power grid and a preset switching state of discrete reactive power regulation equipment, and acquiring an upper training model and a switching action instruction of the discrete reactive power equipment; deploying the lower-layer training model and the upper-layer training model to a power distribution network in actual operation, and obtaining an overall operation model; and constructing a target function according to the operation network loss and the switching action instruction of the discrete reactive power equipment, solving the target function by adopting an A2C algorithm, and obtaining a reactive power optimization result of the power distribution network. According to the invention, the voltage distribution is optimized and the network loss is reduced by obtaining the overall operation model and combining the network loss and the switching action instruction.

Description

technical field [0001] The invention relates to the technical field of distribution network optimization, in particular to a data-driven reactive power optimization method and device for a distribution network. Background technique [0002] With the development of new energy and distributed generation technology, a large number of distributed generation (DG) is connected to the power grid, resulting in an increase in the penetration rate of the distribution network and increased uncertainty. According to statistics, only 51% of the DGs in the 10kV distribution network are connected to the dispatching master station, and the connection rate of three remotes and above is less than 10%. The medium-voltage distribution network presents a high DG penetration rate and low perception characteristics. Traditional centralized reactive power and voltage control is mostly based on swarm intelligence optimization algorithm, which requires complex calculations and requires accurate and c...

AI Technical Summary

Problems solved by technology

[0003] The purpose of the present invention is to provide a data-driven distribution network reactive power optimization method and device to solve the problem in the prior art that the limited measurable node voltage data cannot fully reflect the voltage fluctuation of the target area power grid

Images