基于 Adam-RBF 神经网络的储能 VSG 多参数协同自适应控制策略

杨森; 田桂珍; 刘广忱; 孙冷

引用本文:	杨森,田桂珍,刘广忱,等.基于 Adam-RBF 神经网络的储能 VSG 多参数协同自适应控制策略[J].电力系统保护与控制,2026,54(10):59-70.
	YANG Sen,TIAN Guizhen,LIU Guangchen,et al.A multi-parameter coordinated adaptive control strategy for energy storage VSG based on Adam-RBF neural network[J].Power System Protection and Control,2026,54(10):59-70

【打印本页】【下载PDF全文】【查看/发表评论】【EndNote】【RefMan】【BibTex】

←前一篇|后一篇→

过刊浏览高级检索

本文已被：浏览 129次下载 16次	码上扫一扫！
分享到：微信更多字体:加大+\|默认\|缩小-
基于 Adam-RBF 神经网络的储能 VSG 多参数协同自适应控制策略
杨森¹，田桂珍^1,2，刘广忱^1,2，孙冷¹
1. 内蒙古工业大学电力学院，内蒙古呼和浩特 010080；2. 大规模储能技术教育部工程研究中心，内蒙古呼和浩特 010080

摘要:

为了提升储能虚拟同步发电机 (virtual synchronous generator, VSG) 控制的频率支撑性能，提出了基于适应性矩估计算法的径向基函数 (adaptive moment estimation-radial basis function, Adam-RBF) 神经网络的储能 VSG 多参数协同自适应控制策略。首先，建立风 - 储 - 火联合系统的调频响应模型，推导计及火电和储能 VSG 控制的频率传递函数，定量分析 VSG 的转动惯量、阻尼系数和调频系数对一次调频性能的影响。然后，研究储能 VSG 多参数协调控制策略。该控制策略利用 RBF 神经网络算法来拟合转动惯量、阻尼系数以及调频系数三者之间的非线性关系。同时引入 Adam 算法，显著加快了系统一次调频的恢复速度，减少了权值的迭代次数，降低了对初始参数的依赖性。最后，通过仿真和实验结果验证了所提控制策略能够减小系统频率波动，使系统频率更快达到稳定状态。

关键词: VSG RBF 神经网络算法多参数协同 Adam 算法

DOI：10.19783/j.cnki.pspc.251342

分类号:

基金项目:国家重点研发计划专项资助 (2024YFB2408400)

A multi-parameter coordinated adaptive control strategy for energy storage VSG based on Adam-RBF neural network

YANG Sen^1,2, TIAN Guizhen^1,2, LIU Guangchen^1,2, SUN Leng¹

1. College of Electric Power, Inner Mongolia University of Technology, Hohhot 010080, China; 2. Engineering Research Center of Large-scale Energy Storage Technology of Ministry of Education, Hohhot 010080, China

Abstract:

To enhance the frequency support performance of energy storage systems with virtual synchronous generator (VSG) control, a multi-parameter coordinated adaptive control strategy based on an adaptive moment estimation-radial basis function (Adam-RBF) neural network is proposed. First, a frequency regulation model of a wind-storage-thermal power integrated system is established. The frequency transfer function considering both thermal power units and VSG-controlled energy storage is derived, and the impacts of VSG virtual inertia, damping coefficient, and frequency regulation coefficient on primary frequency regulation performance are quantitatively analyzed. Then, a multi-parameter coordination strategy for energy storage VSG is developed. The strategy employs an RBF neural network to approximate the nonlinear relationships among virtual inertia, damping coefficient, and frequency regulation coefficient. Meanwhile, the Adam algorithm is incorporated to significantly accelerate frequency recovery in primary regulation, reduce the number of weight iterations, and decrease dependence on initial parameter settings. Finally, simulation and experimental results demonstrate that the proposed control strategy effectively suppresses system frequency fluctuations and facilitates faster frequency stabilization.

Key words: VSG RBF neural network algorithm multi-parameter coordination Adam algorithm

X关闭