王萧博,仇晨光,崔玉,周海强,王宇彤
WANG Xiaobo,QIU Chenguang,CUI Yu,ZHOU Haiqiang,WANG Yutong

• 1:国网江苏省电力有限公司电力科学研究院
• 2:国网江苏省电力有限公司
• 3:河海大学电气与动力工程学院

摘要(Abstract)：

现有基于人工神经网络的TSA（暂态稳定评估）方法常采用少数时间断面特征作为输入，无法反映故障演变的时序特征，制约了暂态稳定评估精度。为此，提出了基于TCN-AM（时间卷积网络-注意力机制）及多阶段特征的TSA方法。采集故障前、故障中和故障后的多个时间断面信息构建反映系统动态过程的多阶段特征序列，应用TCN的因果扩张卷积运算有效提取其中蕴含的时序依赖关系；识别受扰严重机组并剔除其余机组特征以实现特征降维，补充少数类样本以提高样本均衡性，并采用改进焦点损失函数以提高模型对不稳定样本的关注；引入多头自注意力机制，自适应地聚焦并增强高价值关键特征。在含风电的IEEE 39节点系统中进行验证，结果表明所提模型具有较好的TSA精度，且模型训练速度快，可解释性好。
Existing transient stability assessment(TSA) methods based on artificial neural networks commonly employ features from only a few time instants as inputs, making them unable to capture the temporal characteristics of fault evolution and thereby limiting the assessment accuracy. To address this issue, a TSA method based on a temporal convolutional network with an attention mechanism(TCN-AM) and multi-stage features is proposed. Information from multiple time instants before, during, and after faults is collected to construct multi-stage feature sequences that characterize the dynamic evolution of power systems. The causal dilated convolution operation of TCN is employed to effectively extract the temporal dependencies embedded in the feature sequences. The severely disturbed generators are identified, while irrelevant generator features are eliminated for feature dimensionality reduction. Minority class samples are supplemented to improve sample balance, and an improved focal loss function is adopted to enhance the model's attention to unstable samples. Furthermore, a multi-head self-attention mechanism is introduced to adaptively focus on and enhance high-value critical features. The proposed method is validated on the IEEE 39-bus system integrated with wind power generation. The results demonstrate that the proposed model achieves high TSA accuracy, fast training speed, and good interpretability.

关键词(KeyWords)： 暂态稳定评估;多阶段特征;时间卷积网络;特征降维;样本均衡性;注意力机制
transient stability assessment;multi-stage features;temporal convolutional network;feature dimensionality reduction;sample balance;attention mechanism

Abstract:

Keywords:

基金项目(Foundation): 国家电网有限公司科技项目（J2024158）

作者(Author): 王萧博,仇晨光,崔玉,周海强,王宇彤
WANG Xiaobo,QIU Chenguang,CUI Yu,ZHOU Haiqiang,WANG Yutong

DOI: 10.19585/j.zjdl.202610001

参考文献(References)：

• [1]张子扬，张宁，杜尔顺，等.双高电力系统频率安全问题评述及其应对措施[J].中国电机工程学报，2022,42(1):1-24.Zhang Ziyang,Zhang Ning,Du Ershun,et al.Review and countermeasures on frequency security issues of power systems with high shares of renewables and power electronics[J].Proceedings of the CSEE,2022,42(1):1-24.
• [2]Odun-Ayo T,Crow M L.Structure-preserved power system transient stability using stochastic energy functions[J].IEEE Transactions on Power Systems,2012,27(3):1450-1458.
• [3]于珍，沈沉，刘锋，等.计及风电不确定性的含暂态稳定约束的经济调度[J].中国电机工程学报，2020,40(22):7270-7282.Yu Zhen,Shen Chen,Liu Feng,et al.Transient stabilityconstrained economic dispatch considering wind power uncertainty[J]. Proceedings of the CSEE,2020,40(22):7270-7282.
• [4]Xia S W,Luo X,Chan K W,et al.Probabilistic transient stability constrained optimal power flow for power systems with multiple correlated uncertain wind generations[J]. IEEE Transactions on Sustainable Energy,2016,7(3):1133-1144.
• [5]Lv H Y,Tang H Y.Machine learning methods and their application research[C]//2011 2nd International Symposium on Intelligence Information Processing and Trusted Computing. October 22-23,2011. Wuhan,China. IEEE,2011:108-110.
• [6]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 newgeneration artificial intelligence technology[J]. Automation of Electric Power Systems,2022,46(22):200-223.
• [7]刘俐，李勇，曹一家，等.基于支持向量机和长短期记忆网络的暂态功角稳定预测方法[J].电力自动化设备，2020,40(2):129-136.Liu Li,Li Yong,Cao Yijia,et al.Transient rotor angle stability prediction method based on SVM and LSTM network[J].Electric Power Automation Equipment,2020,40(2):129-136.
• [8]庞传军，张波，余建明.基于LSTM循环神经网络的短期电力负荷预测[J].电力工程技术，2021,40(1):175-180.Pang Chuanjun,Zhang Bo,Yu Jianming. Short-term power load forecasting based on LSTM recurrent neural network[J]. Electric Power Engineering Technology,2021,40(1):175-180.
• [9]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):2945-2954.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):2945-2954.
• [10]解治军，张东霞，韩肖清，等.基于改进长短期记忆网络的电力系统暂态稳定评估方法研究[J].电网技术，2024,48(3):998-1007.Xie Zhijun,Zhang Dongxia,Han Xiaoqing,et al.Research on transient stability assessment method of power system based on improved long short term memory network[J].Power System Technology,2024,48(3):998-1007.
• [11]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.
• [12]房佳姝，刘崇茹，苏晨博，等.基于自注意力Transformer编码器的多阶段电力系统暂态稳定评估方法[J].中国电机工程学报，2023,43(15):5745-5759.Fang Jiashu,Liu Chongru,Su Chenbo,et al. Multi-stage transient stability assessment of power system based on self-attention transformer encoder[J]. Proceedings of the CSEE,2023,43(15):5745-5759.
• [13]Zhu L P,Lu C,Dong Z Y,et al. Imbalance learning machine-based power system short-term voltage stability assessment[J].IEEE Transactions on Industrial Informatics,2017,13(5):2533-2543.
• [14]田芳，周孝信，史东宇，等.基于卷积神经网络综合模型和稳态特征量的电力系统暂态稳定评估[J].中国电机工程学报，2019,39(14):4025-4032.Tian Fang,Zhou Xiaoxin,Shi Dongyu,et al. Power system transient stability assessment based on comprehensive convolutional neural network model and steady-state features[J].Proceedings of the CSEE,2019,39(14):4025-4032.
• [15]安军，艾士琪，刘道伟，等.基于短时受扰轨迹的电力系统暂态稳定评估方法[J].电网技术，2019,43(5):1690-1697.An Jun,Ai Shiqi,Liu Daowei,et al.A power system transient stability assessment method based on short-time disturbed trajectories[J]. Power System Technology,2019,43(5):1690-1697.
• [16]周生存，罗毅，易煊承，等.基于边界强化混合采样的两阶段电力系统暂态稳定评估[J].电力自动化设备，2024,44(4):143-150.Zhou Shengcun,Luo Yi,Yi Xuancheng,et al.Two-stage transient stability assessment of power system based on boundary enhanced hybrid sampling[J]. Electric Power Automation Equipment,2024,44(4):143-150.
• [17]时纯，刘君，梁卓航，等.基于GAN和多通道CNN的电力系统暂态稳定评估[J].电网技术，2022,46(8):3191-3202.Shi Chun,Liu Jun,Liang Zhuohang,et al.Transient stability assessment of power system based on GAN and multichannel CNN[J]. Power System Technology,2022,46(8):3191-3202.
• [18]胡杰祥，刘君，李岩松，等.基于GAN关键样本增强和迁移学习的暂态稳定自适应评估方法[J/OL].华北电力大学学报，https：//link. Cnki. net/urlid/13.1212.[2025-11-27].Hu Jiexiang,Liu Jun,Li Yansong,et al.Adaptive transient stability assessment method based on Gan key sample enhancement and transfer learning[J/OL].Journal of North China Electric Power University,https：//link. Cnki. net/urlid/13.1212.[2025-11-27].
• [19]肖龙，张靖，何宇，等.基于时间卷积和自适应图卷积网络的电力系统暂态稳定评估[J].电网技术，2025,49(11):4580-4590.Xiao Long,Zhang Jing,He Yu,et al.Power system transient stability assessment based on temporal convolution and adaptive graph convolution network[J]. Power System Technology,2025,49(11):4580-4590.
• [20]Bai S,Kolter J Z,Koltun V.Trellis networks for sequence modeling[C]//7th International Conference on Learning Representations,ICLR 2019,May 6,2019,New Orleans,USA:1-17.
• [21]魏少攀，杨明，韩学山，等.一种利用机端电压幅值轨迹MLE指标的暂态功角稳定在线辨识方法[J].中国电机工程学报，2017,37(13):3775-3786.Wei Shaopan,Yang Ming,Han Xueshan,et al.An online identification method for transient angle stability using MLE index of generator terminal voltage amplitude trajectory[J]. Proceedings of the CSEE,2017,37(13):3775-3786.
• [22]范士雄，赵泽宁，郭剑波，等.数据驱动的电力系统暂态稳定评估方法综述[J].中国电机工程学报，2024,44(9):3408-3428.Fan Shixiong,Zhao Zening,Guo Jianbo,et al.Review on data-driven power system transient stability assessment technology[J]. Proceedings of the CSEE,2024,44(9):3408-3428.
• [23]蔡榕，杨雪，田江，等.基于相关性分析和生成对抗网络的电网缺失数据填补方法[J].电力工程技术，2024,43(1):229-237.Cai Rong,Yang Xue,Tian Jiang,et al. A power system missing data filling method based on correlation analysis and generative adversarial network[J].Electric Power Engineering Technology,2024,43(1):229-237.
• [24]姬凯旋，乔骥，赵紫璇，等.考虑样本分布不均衡的电力系统暂态稳定自适应评估[J].电网技术，2025,49(6):2302-2310.Ji Kaixuan,Qiao Ji,Zhao Zixuan,et al. Adaptive assessment of transient stability of power systems considering unbalanced sample distribution[J]. Power System Technology,2025,49(6):2302-2310.
• [25]陆旦宏，范文尧，杨婷，等.基于生成对抗Transformer的电力负荷数据异常检测[J].电力工程技术，2024,43(1):157-164.Lu Danhong,Fan Wenyao,Yang Ting,et al.Anomaly detection of power load data based on Transformer and generative adversarial networks[J].Electric Power Engineering Technology,2024,43(1):157-164.
• [26]Vaswani A,Shazeer N,Parmar N,et al. Attention is all you need[C]//Proceedings of the 31st International Conference on Neural Information Processing Systems.Curran Associates Inc.,2017:5998-6008.
• [27]杨雨昕，姚伟，邓贤哲，等.基于Koopman时序延拓和CNN-Transformer模型的频率稳定指标预测[J].电网技术，2025,49(8):3474-3485.Yang Yuxin,Yao Wei,Deng Xianzhe,et al.Frequency stability index prediction based on Koopman time series extension and CNN-Transformer model[J]. Power System Technology,2025,49(8):3474-3485.
• [28]张哲，秦博宇，高鑫，等.融合注意力机制和卷积神经网络的电网暂态电压稳定评估及可解释性分析[J].电网技术，2024,48(11):4648-4657.Zhang Zhe,Qin Boyu,Gao Xin,et al. Transient voltage stability assessment and interpretability analysis of power grids based on the fusion of attention mechanism and convolutional neural network[J].Power System Technology,2024,48(11):4648-4657.
• [29]杨凯璇，卢国强，傅国斌，等.基于关键影响因素量化分析和HGNN的暂态稳定评估[J].浙江电力，2025,44(7):33-43.Yang Kaixuan,Lu Guoqiang,Fu Guobin,et al.A transient stability assessment method using quantitative analysis of key influencing factors and HGNN[J]. Zhejiang Electric Power,2025,44(7):33-43.
• [30]郑瀚林，温步瀛，王怀远.融合样本加权迁移学习的暂态电压稳定评估模型[J].浙江电力，2026,45(2):39-46.Zheng Hanlin,Wen Buying,Wang Huaiyuan.A transient voltage stability assessment model integrating sampleweighted transfer learning[J]. Zhejiang Electric Power,2026,45(2):39-46.