基于注意力增强LSTM-GPR的构网型储能变流器电磁暂态模型参数辨识

伍阳阳; 文天舒; 张彩强; 李晓龙; 张伟; 徐恒山

doi:10.19799/j.cnki.2095-4239.2026.0417

您当前的位置：

首页 >

文章列表页 >

基于注意力增强LSTM-GPR的构网型储能变流器电磁暂态模型参数辨识

储能XXXX | 更新时间：2026-07-20

- 基于注意力增强LSTM-GPR的构网型储能变流器电磁暂态模型参数辨识
- Parameter Identification of Electromagnetic Transient Model for Grid-Forming Energy Storage Converters Based on Attention-Enhanced LSTM-GPR
- 储能科学与技术 2026年页码：1-11
- 作者机构：
  
  1.云南电力试验研究院（集团）有限公司，云南昆明 650214
  2.华能新能源股份有限公司云南分公司，云南昆明 650204
  3.三峡大学电气与新能源学院，湖北宜昌 443002
- 作者简介：
  
  伍阳阳（1988—），男，硕士、高级工程师，主要从事新能源发电与并网分析方面的研究。445800890@qq.com
- 基金信息：
  
  云南省新型电力系统技术创新中心(202405AK340007);新能源电力系统全国重点实验室开放课题(LAPS24006)
- DOI：10.19799/j.cnki.2095-4239.2026.0417
  中图分类号： TM 762
- 收稿：2026-05-16，
  
  修回：2026-07-15，
  
  网络首发：2026-07-18，
- 稿件说明：
移动端阅览
伍阳阳, 文天舒, 张彩强, 等. 基于注意力增强LSTM-GPR的构网型储能变流器电磁暂态模型参数辨识[J]. 储能科学与技术, XXXX, XX(XX): 1-11.

WU Yangyang, WEN TianshuXu, YUAN Huayu, et al. Parameter Identification of Electromagnetic Transient Model for Grid-Forming Energy Storage Converters Based on Attention-Enhanced LSTM-GPR[J]. Energy Storage Science and Technology, XXXX, XX(XX): 1-11.
伍阳阳, 文天舒, 张彩强, 等. 基于注意力增强LSTM-GPR的构网型储能变流器电磁暂态模型参数辨识[J]. 储能科学与技术, XXXX, XX(XX): 1-11. DOI： 10.19799/j.cnki.2095-4239.2026.0417.

WU Yangyang, WEN TianshuXu, YUAN Huayu, et al. Parameter Identification of Electromagnetic Transient Model for Grid-Forming Energy Storage Converters Based on Attention-Enhanced LSTM-GPR[J]. Energy Storage Science and Technology, XXXX, XX(XX): 1-11. DOI： 10.19799/j.cnki.2095-4239.2026.0417.

摘要

针对构网型储能控制器参数无法从厂家直接获取而不能对其控制暂态特性进行精准分析的问题，提出基于融合自适应注意力机制(Attention Mechanism Augmented，ATA)的长短期记忆网络（long short-term memory，LSTM）与高斯过程回归（Gaussian Process Regression，GPR）的构网型储能内部特性辨识方法。首先，根据构网型储能控制结构在仿真平台搭建电磁暂态待辨识模型；然后针对待辨识模型设计ATA-LSTM-GPR混合辨识方法，利用LSTM算法结合ATA机制准确提取构网型储能的动态时序特征，采用GPR算法对控制参数进行点估计和不确定性量化以实现构网型储能控制参数的精确辨识。最后，为验证所提方法的正确性和有效性，基于RT-LAB平台搭建构网型储能测试平台并采集动态响应数据集。试验结果表明：所提ATA-LSTM-GPR混合算法能够显著提升构网型储能控制参数的辨识精度，在复杂的高/低电压穿越工况下，辨识算法得到的控制参数具有高保真性和适用性。

Abstract

To address the problem that the control parameters of grid‑forming energy storage converters cannot be directly obtained from manufacturers

thereby impeding accurate analysis of their control transient characteristics

this paper proposes an identification method for the internal characteristics of grid‑forming energy storage converters. The method is based on a long short‑term memory (LSTM) network fused with an adaptive attention mechanism (Attention Mechanism Augmented

ATA) and Gaussian process regression (GPR).First

an electromagnetic transient model to be identified is established on a simulation platform according to the control structure of grid‑forming energy storage converters. Then

a hybrid ATA‑LSTM‑GPR identification method is designed for the model to be identified. The LSTM algorithm combined with the ATA mechanism is used to accurately extract the dynamic time‑series features of grid‑forming energy storage converters

and the GPR algorithm is adopted to perform point estimation and uncertainty quantification of control parameters

so as to realize accurate identification of the control parameters.Finally

to verify the correctness and effectiveness of the proposed method

a grid‑forming energy storage test platform is built based on the RT‑LAB platform

and a dynamic response dataset is collected. Experimental results show that the proposed hybrid ATA‑LSTM‑GPR algorithm can significantly improve the identification accuracy of control parameters for grid‑forming energy storage converters. Under complex high/low voltage ride‑through conditions

the control parameters obtained by the identification algorithm exhibit high fidelity and applicability.

关键词

Keywords

references

Sun P, Tian Z, Huang M, et al.Additional Kinetic Energy Injection and Piecewise Damping Based Postfault Antiwindup and Transient Stability Enhanced Control for Grid-Forming Inverter[J].IEEE Transactions on Power Electronics, 2024(7 Pt.1):39.

马悦鑫,王鹏,赵浩然,等.基于分段仿射阻抗模型的电压源变流器小扰动稳定域在线构造[J].中国电机工程学报,2024,44(14):5667-5680.

MA Y X, WANG P, ZHAO H R, et al. Online construction of small-signal stability region for voltage source converter based on piecewise affine impedance model[J]. Proceedings of the CSEE, 2024, 44(14): 5667-5680.

胡朝华,杜新伟,苟竞,等.提升新型电力系统宽频振荡稳定性的构网型储能设备优化配置方法[J].储能科学与技术,2025,14(12):4583-4593.

HU C H, DU X W, GOU J, et al. Optimal configuration method of grid-forming energy storage devices for improving broadband oscillation stability of new power systems[J]. Energy Storage Science and Technology, 2025, 14(12): 4583-4593.

颜湘武,张书瑞,蔡光,等.构网型双馈风电机组稳定性分析[J].太阳能学报,2025,46(10):561-569.

YAN X W, ZHANG S R, CAI G, et al. Stability analysis of grid-forming doubly-fed wind turbines[J]. Acta Energiae Solaris Sinica, 2025, 46(10): 561-569.

Ai C , Li Y , Zhao Z ,et al.An Extension of Grid-Forming: A Frequency-Following Voltage-Forming Inverter[J].IEEE Transactions on Power Electronics, 2024(10 Pt.1):39.

张梦琪,李永刚,孙庚,等.基于DBN-ELM的构网型并网逆变器控制参数自适应调整方法[J].电力自动化设备,2024,44(04):111-118.

ZHANG M Q, LI Y G, SUN G, et al. Adaptive adjustment method of control parameters for grid-forming grid-connected inverter based on DBN-ELM[J]. Electric Power Automation Equipment, 2024, 44(4): 111-118.

秦继朔,贾科,孔繁哲,等. 基于寻优算法的永磁风机并网逆变器故障穿越控制参数分步辨识[J].中国电机工程学报, 2021, 41(S1): 59-69.

QIN J S, JIA K, KONG F Z, et al. Step-by-step identification of fault ride-through control parameters for permanent magnet wind turbine grid-connected inverter based on optimization algorithm[J]. Proceedings of the CSEE, 2021, 41(S1): 59-69.

王渝红,王雪珂,文玉玲,等.基于MCC-PCKF的跟网型变流器时变虚拟惯量与状态估计[J].电力系统自动化,2025,49(13):103-112.

WANG Y H, WANG X K, WEN Y L, et al. Time-varying virtual inertia and state estimation of grid-following converter based on MCC-PCKF[J]. Automation of Electric Power Systems, 2025, 49(13): 103-112.

M. Chen, D. Zhou, and F. Blaabjerg, "Enhanced Virtual Synchronous Generator with Adaptive Inertia and Damping Support," IEEE Transactions on Industrial Electronics, vol. 70 , no. 6 , pp. 5845 - 5856 , Jun. 2023 .

Mahmood Z I , Cui H , She B ,et al.Evaluating the Equivalent Inertia of Grid-Following and Grid-Forming Inverter-Based Resources[J].IEEE Transactions on Energy Conversion, 2024:1-12.

S. Hochreiter and J. Schmidhuber, "Long Short-Term Memory," Neural Computation, vol. 9 , no. 8 , pp. 1735 - 1780 , 1997 .

李毅丰,裴玮,熊佳旺,等.构网型变流器的功率-相角直接控制策略[J].中国电机工程学报,2025,45(S1):71-82.

LI Y F, PEI W, XIONG J W, et al. Direct power-angle control strategy for grid-forming converters[J]. Proceedings of the CSEE, 2025, 45(S1): 71-82.

宾子君,孔祥平,黄云辉,等.接入交流电网的构网型储能变流器阻抗建模及稳定性分析[J].储能科学与技术,2025,14(9):3434-3443..

BIN Z J, KONG X P, HUANG Y H, et al. Impedance modeling and stability analysis of grid-forming energy storage converters connected to AC grid[J]. Energy Storage Science and Technology, 2025, 14(9): 3434-3443.

李明烜,王跃,徐宁一,等. 基于带通阻尼功率反馈的虚拟同步发电机控制策略[J]. 电工技术学报, 2018, 33(10): 2176-2185.

LI M X, WANG Y, XU N Y, et al. Control strategy of virtual synchronous generator based on band-pass damping power feedback[J]. Transactions of China Electrotechnical Society, 2018, 33(10): 2176-2185.

Quan X , Li D , Zou Z ,et al.Complex Variable Design for Power Control of Grid-Forming Inverter[J].IEEE Transactions on Sustainable Energy, 2025:1-4.

Sun L , Ding G , Geng S .A New Neural Network Model Based on Attention Mechanism that Embeds LSTM into RNN for Nonlinear Time-Lag System Identification[J].Lecture Notes in Electrical Engineering, 2024:610-618.

Yang R , Peng S , Yao G .Multimodal feature enhanced Bi-LSTM model for harmonic power load identification in distribution networks[J].Journal of Power Electronics, 2025, 25(6):1116-1126.

施永,黄宁,谢缔,等.基于自注意力机制和CNN融合的燃料电池故障诊断技术[J].太阳能学报,2025,46(05):53-61.

SHI Y, HUANG N, XIE D, et al. Fault diagnosis technology for fuel cells based on fusion of self-attention mechanism and CNN[J]. Acta Energiae Solaris Sinica, 2025, 46(5): 53-61.

褚昱麟,陈红坤,王嘉琦,等.基于多元高斯过程回归和空间聚类的复信号谐波源建模[J].电工技术学报,2025,40(23):7621-7634.

CHU Y L, CHEN H K, WANG J Q, et al. Modeling of complex signal harmonic sources based on multivariate Gaussian process regression and spatial clustering[J]. Transactions of China Electrotechnical Society, 2025, 40(23): 7621-7634.

徐恒山,李颜汝,李文昊,等.融合寻优算法的双馈风力机控制参数分步辨识方法[J].太阳能学报,2024,45(04):247-256.

XU H S, LI Y R, LI W H, et al. Step-by-step identification method of control parameters for doubly-fed wind turbine integrating optimization algorithm[J]. Acta Energiae Solaris Sinica, 2024, 45(4): 247-256.

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

基于双层机器学习的非对称超级电容器跨表征EIS参数估计

基于热电联合模型和深度学习的锂离子电池温度估计方法

基于单颗粒电化学模型的多种滥用情况下锂电内短路程度评估

非高斯噪声条件下的锂电池鲁棒SoE估计

基于白鹭群优化高斯过程回归的锂电池SOH估计方法