Smart power grid partition network reconstruction method based on multi-agent reinforcement learning

Abstract

The invention provides an intelligent power grid partition network reconstruction method based on multi-agent reinforcement learning, and the method comprises the following steps: 1, dividing a power grid into N regions according to the operation demands of the power grid, and constructing the basic elements of the multi-agent reinforcement learning, including environment, agents, states, observation, actions and reward functions; 2, operating the simulation environment of the power system, and creating an initial operation state data set of the power system; 3, constructing a deep neural network model, and training a decision-making agent by applying learning between enhanced agents; and 4, providing a strategy for power grid reconstruction by using the trained intelligent agent. According to the invention, through interaction between multiple agents and a power simulation environment, an optimal network reconstruction strategy is learned offline and is applied to an actual power grid online.

Description

technical field [0001] The invention relates to the field of multi-agent reinforcement learning, in particular to a smart grid partition network reconfiguration method based on multi-agent reinforcement learning Background technique [0002] Network reconfiguration refers to changing the network topology of the power grid, that is, changing the operating state of the tie switch and section switch of the power grid, so as to transfer the load between feeders or distribution stations, thereby changing the operating state of the power grid. When the power grid fails, network reconfiguration can restore the safe and stable operation of the power grid. Traditional network reconfiguration relies on optimization algorithms or expert experience. Optimization algorithms often have a large amount of calculation and slow processing speed, which is not conducive to real-time applications. Expert experience lacks the means to deal with possible risks that have not occurred, and it is di...

AI Technical Summary

Problems solved by technology

Traditional network reconfiguration relies on optimization algorithms or expert experience. Optimization algorithms often have a large amount of calculation and slow processing speed, which is not conducive to real-time applications

Expert experience lacks the means to deal with possible risks that have never occurred, and it is difficult to solve the increasingly complex power system operation safety issues

In addition, it is difficult to consider the uncertainty of wind power, photovoltaic power generation and load at the same time in traditional network reconfiguration

Before performing network reconfiguration, it is necessary to estimate the operating state of the power grid after network reconfiguration. The accuracy of the estimation directly determines the quality of network reconfiguration actions, which increases the difficulty of network reconfiguration.

Images