基于物理信息嵌入的非固定长度电力系统暂态稳定快速评估

doi:10.16183/j.cnki.jsjtu.2023.452

摘要/Abstract

摘要：

在双碳目标下,构建以新能源为主体的新型电力系统是实现电力工业转型升级的主要方向和关键途径,新型电力系统背景下快速准确的暂态功角稳定评估研究具有重要意义.为此,基于物理信息嵌入序列到序列(PI-seq2seq)神经网络与级联卷积神经网络模型提出一种含构网型新能源的新型电力系统暂态功角稳定评估方法.首先,采用PI-seq2seq网络结构预测未来功角轨迹,通过构造含物理损失项的损失函数引导模型训练过程,避免时域仿真耗时过长影响快速暂态评估.其次,级联卷积神经网络以预测的功角轨迹作为输入评估暂态稳定情况及其置信度,并配置评估置信度阈值判断机制以实现非固定评估长度的暂态稳定判断,克服了固定功角曲线长度对评估结果的影响.最后,在Kundur系统中进行验证,仿真结果表明:所提方法在功角曲线预测与稳定评估方法均获得令人满意的结果.

关键词: 构网型新能源, 物理信息嵌入序列到序列神经网络, 功角轨迹预测, 级联卷积神经网络, 暂态功角稳定评估

Abstract:

Against the backdrop of “dual carbon” goals, the construction of a new power system with new energy as the main component is the main direction and key way for the transformation and upgrading of the power industry. Research into fast and accurate evaluation of transient power angle stability in the context of new power systems is of great significance. To address this, a new transient power angle stability evaluation method is proposed for power systems with grid-forming new energy based on the physics-informed sequence-to-sequence (PI-seq2seq) neural networks and cascaded convolutional neural networks models. First, the PI-seq2seq network structure is used to predict the future power angle trajectory, and a loss function with physical loss terms is constructed to guide the model training process, which avoids the long-duration time-domain simulation to ensure fast transient evaluation. Then, predicted power angle trajectory is taken as input by the cascade convolutional neural networks to evaluate the transient stability and its confidence level. A threshold judgment mechanism for the evaluation confidence level is configured to realize the transient stability judgment of the non-fixed evaluation length, which overcomes the impact of the fixed power angle curve length on the evaluation results. Finally, the method proposed is verified in the Kundur system, and the simulation results show that it has obtained satisfactory results in both the power angle curve prediction and the stability evaluation.

Key words: grid-forming new energy, physics-informed sequence-to-sequence (PI-seq2seq) neural networks, power angle trajectory prediction, cascade convolutional neural networks, transient power angle stability assessment

中图分类号:

TM712
TP183

李湘, 陈思远, 张俊, 柯德平, 高杰迈, 杨欢欢. 基于物理信息嵌入的非固定长度电力系统暂态稳定快速评估[J]. 上海交通大学学报, 2025, 59(7): 962-970.

LI Xiang, CHEN Siyuan, ZHANG Jun, KE Deping, GAO Jiemai, YANG Huanhuan. Physics-Informed Fast Transient Stability Assessment of Non-Fixed Length in Power Systems[J]. Journal of Shanghai Jiao Tong University, 2025, 59(7): 962-970.

图/表 9

图1

表1

图2

表2

图3

表3

表4

表5

表6

参考文献 23

[1]	中华人民共和国国家能源局. 国家能源局2023年一季度新闻发布会文字实录[EB/OL]. (2023-02-13)[2023-8-28]. http://www.nea.gov.cn/2023-02/13/c_1310697149.htm.
	National Energy Administration of the People’s Republic of China. Transcript of the National Energy Administration’s press conference for the first quarter of 2023[EB/OL]. (2023-02-13)[2023-8-28]. http://www.nea.gov.cn/2023-02/13/c_1310697149.htm.
[2]	何剑, 屠竞哲, 孙为民, 等. 美国加州“8·14” 、“8·15” 停电事件初步分析及启示[J]. 电网技术, 2020, 44(12): 4471-4478.
	HE Jian, TU Jingzhe, SUN Weimin, et al. Preliminary analysis and lessons of California power outage events on August 14 and 15, 2020[J]. Power System Technology, 2020, 44(12): 4471-4478.
[3]	樊陈, 姚建国, 张琦兵, 等. 英国“8·9” 大停电事故振荡事件分析及思考[J]. 电力工程技术, 2020, 39(4): 34-41.
	FAN Chen, YAO Jianguo, ZHANG Qibing, et al. Reflection and analysis for oscillation of the blackout event of 9 August 2019 in UK[J]. Electric Power Engineering Technology, 2020, 39(4): 34-41.
[4]	王伟胜, 林伟芳, 何国庆, 等. 美国得州2021年大停电事故对我国新能源发展的启示[J]. 中国电机工程学报, 2021, 41(12): 4033-4043.
	WANG Weisheng, LIN Weifang, HE Guoqing, et al. Enlightenment of 2021 texas blackout to the renewable energy development in China[J]. Proceedings of the CSEE, 2021, 41(12): 4033-4043.
[5]	肖先勇, 郑子萱. “双碳” 目标下新能源为主体的新型电力系统:贡献、关键技术与挑战[J]. 工程科学与技术, 2022, 54(1): 47-59.
	XIAO Xianyong, ZHENG Zixuan. New power systems dominated by renewable energy towards the goal of emission peak & carbon neutrality: Contribution, key techniques, and challenges[J]. Advanced Engineering Sciences, 2022, 54(1): 47-59.
[6]	舒印彪, 陈国平, 贺静波, 等. 构建以新能源为主体的新型电力系统框架研究[J]. 中国工程科学, 2021, 23(6): 61-69. doi: 10.15302/J-SSCAE-2021.06.003
	SHU Yinbiao, CHEN Guoping, HE Jingbo, et al. Building a new electric power system based on new energy sources[J]. Strategic Study of CAE, 2021, 23(6): 61-69. doi: 10.15302/J-SSCAE-2021.06.003
[7]	杨博, 陈义军, 姚伟, 等. 基于新一代人工智能技术的电力系统稳定评估与决策综述[J]. 电力系统自动化, 2022, 46(22): 200-223.
	YANG Bo, CHEN Yijun, YAO Wei, et al. Review on stability assessment and decision for power systems based on new-generation artificial intelligence technology[J]. Automation of Electric Power Systems, 2022, 46(22): 200-223.
[8]	赵恺, 石立宝. 基于改进一维卷积神经网络的电力系统暂态稳定评估[J]. 电网技术, 2021, 45(8): 2945-2957.
	ZHAO Kai, SHI Libao. Transient stability assessment of power system based on improved one-dimensional convolutional neural network[J]. Power System Technology, 2021, 45(8): 2945-2957.
[9]	张亮, 安军, 周毅博. 基于时间卷积和图注意力网络的电力系统暂态稳定评估[J]. 电力系统自动化, 2023, 47(7): 114-122.
	ZHANG Liang, AN Jun, ZHOU Yibo. Transient stability assessment of power system based on temporal convolution and graph attention network[J]. Automation of Electric Power Systems, 2023, 47(7): 114-122.
[10]	李志兵, 肖健梅, 王锡淮. 基于多粒度NRS和改进Bi-LSTM的电力系统暂态稳定评估[J]. 电气工程学报, 2023, 18(3): 232-241.
	LI Zhibing, XIAO Jianmei, WANG Xihuai. Transient stability assessment of power system based on multi-granularity neighborhood rough set and improved bi-directional long-short-term memory network[J]. Journal of Electrical Engineering, 2023, 18(3): 232-241.
[11]	ZHOU Y Z, GUO Q L, SUN H B, et al. A novel data-driven approach for transient stability prediction of power systems considering the operational variability[J]. International Journal of Electrical Power & Energy Systems, 2019, 107: 379-394.
[12]	RAISSI M, PERDIKARIS P, KARNIADAKIS G E. Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations[J]. Journal of Computational Physics, 2019, 378: 686-707.
[13]	MISYRIS G S, VENZKE A, CHATZIVASILEIADIS S. Physics-informed neural networks for power systems[C]// 2020 IEEE Power & Energy Society General Meeting. Montreal, Canada: IEEE, 2020: 1-5.
[14]	MANSOURLAKOURAJ M, GAUTAM M, HOSSAIN R, et al. Event classification in active distribution grids using physics-informed graph neural network and PMU measurements[C]// 2022 IEEE Industry Applications Society Annual Meeting. Detroit, USA: IEEE, 2022: 1-6.
[15]	PLANT R, BABAZADEH D, STOCK S, et al. Real-time inertia estimation in an inverter-dominated distribution grid using a physics-informed recurrent neural network[C]// CIRED Porto Workshop 2022:E-mobility and power distribution systems. Porto, Portugal: IET, 2022: 940-944.
[16]	陈颖, 高仕林, 宋炎侃, 等. 面向新型电力系统的高性能电磁暂态云仿真技术[J]. 中国电机工程学报, 2022, 42(8): 2854-2864.
	CHEN Ying, GAO Shilin, SONG Yankan, et al. High-performance electromagnetic transient simulation for new-type power system based on cloud computing[J]. Proceedings of the CSEE, 2022, 42(8): 2854-2864.
[17]	YU J J Q, HILL D J, LAM A Y S, et al. Intelligent time-adaptive transient stability assessment system[J]. IEEE Transactions on Power Systems, 2018, 33(1): 1049-1058.
[18]	CHEN Q F, WANG H Y. Time-adaptive transient stability assessment based on gated recurrent unit[J]. International Journal of Electrical Power & Energy Systems, 2021, 133: 107156.
[19]	张若愚. 基于卷积神经网络的电力系统暂态稳定评估[D]. 北京: 北京交通大学, 2020.
	ZHANG Ruoyu. Power system transient stability assessment based on convolutional neural networks[D]. Beijing: Beijing Jiaotong University, 2020.
[20]	李福成, 徐箭, 廖思阳, 等. 基于样本关注度和多层次特征的多阶段电力系统暂态稳定评估[J]. 中国电机工程学报, 2021, 41(22): 7596-7608.
	LI Fucheng, XU Jian, LIAO Siyang, et al. Multi-stage power system transient stability assessment based on sample attention and hierarchical features[J]. Proceedings of the CSEE, 2021, 41(22): 7596-7608.
[21]	ZHANG H B, XIANG W, LIN W X, et al. Grid forming converters in renewable energy sources dominated power grid: Control strategy, stability, application, and challenges[J]. Journal of Modern Power Systems & Clean Energy, 2021, 9(6): 1239-1256.
[22]	中华人民共和国国家市场监督管理总局, 国家标准化管理委员会. 电力系统安全稳定导则: GB 38755—2019[S]. 北京: 中国标准出版社, 2019.
	State Administration for Market Regulation, Standardization Administration of the People’s Republic of China. Code on security and stability for power system: GB 38755—2019[S]. Beijing: Standards Press of China, 2019.
[23]	CUI H T, LI F X, TOMSOVIC K. Hybrid symbolic-numeric framework for power system modeling and analysis[J]. IEEE Transactions on Power Systems, 2021, 36(2): 1373-1384.

参数	含义	取值
Input_len	输入长度	5步长
Output_len	预测长度	20步长
Input_dim	输入特征数量	20
Output_dim	输出特征数量	2
Learning_rate	学习率	3×10^-4(预训练)/ 1×10^-4(微调)
Hidden_size	隐藏层大小	128
Epoch	训练轮次	80
Batch_size	训练批次大小	64
Optimizer	优化器	Adam
Loss function	损失函数	RMSE

模型序号	模型	平均绝对误差	均方误差	平均相对误差/%
1	seq2seq	5.531×10^-3	8.412×10^-5	0.7226
2	PI-seq2seq(Euler)	5.199×10^-3	8.975×10^-5	0.4820
3	PI-seq2seq(RK4)	4.359×10^-3	7.117×10^-5	0.4189

计算长度	Andes时域仿真平均耗时/s	PI-seq2seq(RK4) 平均耗时/s
0.3 s功角曲线	0.3480	0.0199
0.5 s功角曲线	0.5300	0.0215

参数	含义	取值
Kernal size	卷积核大小	3×3
Learning_rate	学习率	3×10^-4
Epoch	训练轮次	100
Batch_size	训练批次大小	64
Optimizer	优化器	Adam
Loss function	损失函数	交叉熵损失

层级	准确率	查准率	查全率	F₁
CNN1	97.28	90.74	95.14	92.89
CNN2	98.73	95.23	97.09	96.15
CNN3	99.27	98.05	98.05	98.05