基于Bi-LSTM循环神经网络的风储系统控制策略

Control Strategy for Combined Wind-storage System Based on Bi-LSTM Recurrent Neural Network

“双碳”背景下风电的渗透率不断提高,将对电力系统的形态和运行机制产生深刻影响。本文提出了一种基于双向长短期记忆Bi-LSTM(bidirectional long short-term memory)循环神经网络的风储系统控制策略。采用双向长短时循环神经网络提取控制结果与风电场实际出力以及储能状态间的时序信息,通过构建基于双向长短时记忆循环神经网络的控制模型,使得风电场在多种运行工况下能够快速、准确地得到储能系统调节结果。基于实际风电场数据仿真结果表明,本文所提控制策略能够保证在一定经济效益的前提下,将风储系统控制误差保持在0.50%~1.37%。

The increasing permeability of wind power under the“Dual carbon goals”background will have a profound impact on the form and operation mechanism of power system.In this paper,a control strategy for a combined wind-storage system based on bidirectional long short-term memory(Bi-LSTM)recurrent neural network was proposed.The time sequence information between the control results and the actual output from a wind farm and energy storage state was extracted by the Bi-LSTM recurrent neural network.By constructing a control model based on the Bi-LSTM recurrent neural network,the regulation results of the energy storage system can be obtained quickly and accurately under various operating conditions of the wind farm.Finally,the simulation results based on the data of an actual wind farm show that the proposed control strategy can keep the control error of the combined wind-storage system between 0.50%and 1.37%under the premise of certain economic benefits.