Electric power system source-load look-ahead scheduling method and device based on deep reinforcement learning

Abstract

The invention provides an electric power system source-load look-ahead scheduling method and device based on deep reinforcement learning. The method comprises the following steps: obtaining the economic operation basic data of an electric power system, constructing an electric power system source-load look-ahead scheduling model according to the economic operation basic data of the electric power system, and constructing an electric power system look-ahead scheduling model containing demand side response; designing a state space, an action space and a reward function based on the prospective scheduling model of the power system so as to design a time sequence decision-making mechanism of the economic scheduling problem of the power system; and according to the time sequence decision mechanism, applying a deep reinforcement learning algorithm to the electric power system look-ahead scheduling model, and improving and applying the deep reinforcement learning algorithm to obtain a look-ahead scheduling strategy based on deep reinforcement learning. A solution is provided for intelligent power grid economic optimization dispatching with sufficient supply and demand interaction, participation of a large number of subjects and uncertainty improvement, and the decision-making speed, the reliability and the automation and intelligence level of power system dispatching are improved.

Description

technical field [0001] The invention relates to the technical field of power system optimization scheduling and reinforcement learning, in particular to a power system source-load forward scheduling method and device based on deep reinforcement learning. Background technique [0002] With the gradual advancement of my country's new power system construction, the traditional power grid is gradually developing into a complex power system involving a large number of subjects, and the strengthening of source-load interaction has significantly increased the number of subjects participating in the operation of the power system. In addition, the year-by-year increase in the penetration rate of new energy sources has also brought certain uncertainties to the operation of the power system, increasing the difficulty of optimizing the operation of the power system. The traditional artificial day-ahead scheduling method is difficult to adapt to this new change, and more flexible and eff...

AI Technical Summary

Problems solved by technology

[0004] Existing research on applying reinforcement learning to issues related to power systems mainly focuses on the above three aspects, and there is no literature on applying reinforcement learning to forward-looking optimal dispatching of power systems

Images