Virtual power plant economic dispatching method based on scenes and deep reinforcement learning

Abstract

The invention discloses a virtual power plant economic dispatching method based on scenes and deep reinforcement learning, solves the problem of poor universality of an existing virtual power plant economic dispatching method, and belongs to the field of intelligent power grid economic dispatching. The method comprises the steps: S1, establishing a target function and corresponding constraint conditions of virtual power plant economic dispatch, so as to enable a power system to operate normally; s2, classifying the historical data into data of an extreme scene and data of a normal scene according to a set threshold value; s3, respectively expanding the data of the extreme scene and the data of the normal scene by using a generative adversarial network GAN to obtain an extreme scene data set and a normal scene data set; and S4, training the neural network by using a depth deterministic strategy gradient algorithm and taking the extreme scene data set and the normal scene data set as training sets, solving parameters of the target function, and determining an economic dispatching strategy of the virtual power plant. A virtual power plant (VPP) with energy storage and access to a power distribution network is operated stably under uncertain conditions.

Description

technical field [0001] The invention relates to a virtual power plant economic scheduling method based on scenarios and deep reinforcement learning, belonging to the field of smart grid economic scheduling. Background technique [0002] With the access of a large number of distributed new energy generators in traditional power facilities, it poses challenges to the normal operation of traditional power grids. This is because compared with traditional thermal power generators, new energy generators do not have the properties of traditional thermal power generation: such as continuity, stability, etc. New energy is more dependent on weather in the process of output. And natural conditions, then new energy output has the disadvantages of intermittent, uncertain and so on. However, compared with traditional energy sources, new energy sources also have the characteristics of low cost and less environmental pollution. Due to the above reasons, a new power plant model virtual pow...

AI Technical Summary

Problems solved by technology

[0006] Aiming at the problem of poor versatility of existing virtual power plant economic scheduling methods, the present inv

Images