柔性直流输电系统保护测试用故障暂态波形智能生成方法

Intelligent Generation Method for Fault Transient Waveforms for Protection Testing of Flexible DC Transmission Systems

摘要

柔性直流输电系统保护一般基于故障后的暂态电气量进行区内外故障识别并确定是否跳闸，如何生成能够反映各种故障情况的柔直输电系统直流侧故障暂态电气量波形，是进行保护出厂测试以及阶段性检验的基础．然而，现有的3种故障暂态波形的获取方法存在实用性有限、经济性较差、通用性不强的缺点，并且无法结合柔直保护测试流程进行保护测试用的故障暂态波形批量生成．针对如何快速批量获取柔直保护测试用暂态波形的问题，提出基于频带能量占比的波形暂态特征稀疏频域提取方法，以能量值量化不同频率分量对波形特征的贡献度，通过高能量占比的稀疏频域减少了频域参数量，并避免了低能量占比的频率分量带来的拟合误差；提出了稀疏频域参数计算方法，以各个频率下的幅值与相位参数表征波形的暂态特征，实现故障暂态特征从物理空间至数学空间的显式映射；提出了基于深层神经网络(DNN)的系统和故障参数空间与稀疏频域参数空间的映射关系拟合方法，利用 DNN 的强非线性映射能力实现两个空间的快速映射，进而实现暂态波形的快速智能批量生成．算例结果表明，生成的故障暂态波形与数字仿真波形具有暂态特征一致性．该方法提供了新的柔直输电系统暂态波形生成思路，能够显著降低波形获取的人力、物力与时间成本．

Abstract

Protection of flexible direct current(DC)transmission systems generally relies on postfault transient electrical quantities to identify internal and external faults and determine tripping decisions. Generating DC-side fault transient waveforms that accurately reflect various fault conditions is essential for factory testing and periodic verification of protection devices. However，the existing three methods for generating transient waveforms suffer from limitations：limited practicality，poor cost-effectiveness，and insufficient universality，along with the inability to generate transient waveforms in batches that are integrated with flexible DC protection testing workflows. To address the challenge of rapid batch generation of transient waveforms for protection testing，this study proposes a sparse frequency-domain method based on frequency-band energy contribution ratios. This method quantifies the impact of different frequency components on transient waveform characteristics through energy metrics. By retaining high-energy frequency components and discarding low-energy ones，this method reduces parameter dimensionality and minimizes fitting errors. A sparse frequency-domain parameter calculation method is further developed to explicitly map transient features from physical space to mathematical space using amplitude and phase parameters at key frequencies. Additionally，a deep neural network(DNN)-based framework is established to learn the mapping from parameter space of the system and faults to the sparse frequency-domain parameter space. This framework leverages the nonlinear mapping capabilities of DNN for intelligent batch transient waveform generation. Case study results demonstrate that the generated waveforms exhibit transient characteristics that are consistent with detailed digital simulations. The proposed method introduces a novel paradigm for transient waveform generation in flexible DC transmission systems，considerably reducing the labor，material resources，and time costs associated with traditional waveform generation methods.

关键词

继电保护 / 保护测试 / 人工智能 / 故障暂态波形 / 柔性直流输电系统

Key words

relay protection / protection testing / artificial intelligence / fault transient waveform / flexible direct current(DC) transmission system

参考文献

[1] Wang Y Z, Yuan Z C, Fu J, et al. A feasible coordination protection strategy for MMC—MTDC systems under DC faults[J]. International Journal of Electrical Power & Energy Systems, 2017, 90:103-111.

[2] 刘剑, 邰能灵, 范春菊, 等. 柔性直流输电线路故障处理与保护技术评述[J]. 电力系统自动化, 2015, 39(20):158-167.

[3] Liu Jian, Tai Nengling, Fan Chunju, et al. Comments on fault handling and protection technology for VSC—HVDC transmission lines[J]. Automation of Electric Power Systems, 2015, 39(20):158-167(in Chinese).

[4] 蒋原, 刘谨华, 陈明轩. 柔性直流输电的故障分析及保护策略综述[J]. 化工自动化及仪表, 2022, 49(6):677-685.

[5] Jiang Yuan, Liu Jinhua, Chen Mingxuan. Overview of fault analysis and protection strategies for VSC—HVDC[J]. Control and Instruments in Chemical Industry, 2022, 49(6):677-685(in Chinese).

[6] 杨子千, 马锐, 程时杰, 等. 电力电子化电力系统稳定的问题及挑战：以暂态稳定比较为例[J]. 物理学报, 2020, 69(8):103-116.

[7] Yang Ziqian, Ma Rui, Cheng Shijie, et al. Problems and challenges of power—electronic—based power system stability：A case study of transient stability comparison[J]. Acta Physica Sinica, 2020, 69(8):103-116(in Chinese).

[8] 胡家兵, 袁小明, 程时杰. 电力电子并网装备多尺度切换控制与电力电子化电力系统多尺度暂态问题[J]. 中国电机工程学报, 2019, 39(18):5457-5467，5594.

[9] Hu Jiabing, Yuan Xiaoming, Cheng Shijie. Multi—time scale transients in power—electronized power systems considering multi—time scale switching control schemes of power electronics apparatus[J]. Proceedings of the CSEE, 2019, 39(18):5457-5467，5594(in Chinese).

[10] 贺之渊, 刘栋, 庞辉. 柔性直流与直流电网仿真技术研究[J]. 电网技术, 2018, 42(1):1-12.

[11] He Zhiyuan, Liu Dong, Pang Hui. Research on simulation technologies of VSC—HVDC and DC grids[J]. Power System Technology, 2018, 42(1):1-12(in Chinese).

[12] 刘清泉, 杨鹏, 李铁成, 等. 应对继电保护多场景的数字仿真测试方法[J]. 中国电力, 2024, 58(2):118-125.

[13] Liu Qingquan, Yang Peng, Li Tiecheng, et al. Digital simulation test method for multiple scenarios relay protection[J]. Electric Power, 2024, 58(2):118-125(in Chinese).

[14] 罗震, 董建伟, 胡建明. 基于改进麻雀搜索算法优化 BPNN 的电阻点焊质量预测[J]. 天津大学学报(自然科学与工程技术版), 2024, 57(5):445-451.

[15] Luo Zhen, Dong Jianwei, Hu Jianming. Optimization of resistance spot welding quality prediction based on improved sparrow search algorithm for BPNN[J]. Journal of Tianjin University(Science and Technology), 2024, 57(5):445-451(in Chinese).

[16] 董靖川, 武晓鑫, 高宇博, 等. 融合注意力机制的刀具磨损预测方法[J]. 天津大学学报(自然科学与工程技术版), 2024, 57(4):362-373.

[17] Dong Jingchuan, Wu Xiaoxin, Gao Yubo, et al. Tool wear prediction method with attention mechanism[J]. Journal of Tianjin University(Science and Technology), 2024, 57(4):362-373(in Chinese).

[18] 王晓玲, 梁羽翎, 王佳俊, 等. 耦合注意力机制大坝变形改进 LSTM 序列到序列预测模型[J]. 天津大学学报(自然科学与工程技术版), 2023, 56(7):702-712.

[19] Wang Xiaoling, Liang Yuling, Wang Jiajun, et al. Improved LSTM sequence—to—sequence prediction model for dam deformation coupled with attention mechanism[J]. Journal of Tianjin University(Science and Technology), 2023, 56(7):702-712(in Chinese).

[20] 乔婉淇, 丁建文, 郭强亮, 等. 基于DNN超参数优化的5G—R宽带集群通信故障识别[J]. 铁道科学与工程学报, 2025, 22(10):4749-4760.

[21] Qiao Wanqi, Ding Jianwen, Guo Qiangliang, et al. 5G—R broadband trunking communication fault identification based on DNN hyperparameter optimization[J]. Journal of Railway Science and Engineering, 2025, 22(10):4749-4760(in Chinese).

[22] 曾西平, 刘牧远. 基于VOF法和改进DNN模型的三维液滴曲率算法[J]. 四川轻化工大学学报(自然科学版), 2025, 38(2):37-46.

[23] Zeng Xiping, Liu Muyuan. Three—dimensional droplet curvature algorithm based on VOF method and improved DNN model[J]. Journal of Sichuan University of Science and Engineering(Natural Science Edition), 2025, 38(2):37-46(in Chinese).

[24] 余凯峰, 吐松江·卡日, 张紫薇, 等. 基于级联 MCNN—MMLP 双残差网络的短期负荷预测[J]. 电力系统保护与控制, 2025, 53(2):151-162.

[25] Yu Kaifeng, Tusongjiang·Kari, Zhang Ziwei, et al. Short—term load forecasting based on a cascade MCNN—MMLP dual residual network[J]. Power System Protection and Control, 2025, 53(2):151-162(in Chinese).

[26] 于越, 葛磊蛟, 金朝阳, 等. 考虑天气特征与多变量相关性的配电网短期负荷预测[J]. 电力系统保护与控制, 2024, 52(6):131-141.

[27] Yu Yue, Ge Leijiao, Jin Zhaoyang, et al. Short—term load prediction method of distribution networks considering weather features and multivariate correlations[J]. Power System Protection and Control, 2024, 52(6):131-141(in Chinese).

[28] 习伟, 蔡田田, 张镇, 等. 融合多元影响因素的配电台区 BiLSTM 负荷预测方法[J]. 天津大学学报(自然科学与工程技术版), 2023, 56(11):1205-1216.

[29] Xi Wei, Cai Tiantian, Zhang Zhen, et al. BiLSTM load forecasting method for transformer districts integrated with multiple influencing factors[J]. Journal of Tianjin University(Science and Technology), 2023, 56(11):1205-1216(in Chinese).

[30] 张静, 熊国江. 考虑季节特性与数据窗口的短期光伏功率预测组合模型[J]. 电力工程技术, 2025, 44(1):183-192.

[31] Zhang Jing, Xiong Guojiang. Short—term photovoltaic power prediction combination model considering seasonal characteristic and data window[J]. Electric Power Engineering Technology, 2025, 44(1):183-192(in Chinese).

[32] 张家安, 李凤贤, 王铁成, 等. 一种采用记忆神经网络和曲线形状修正的负荷预测方法[J]. 电力工程技术, 2024, 43(1):117-126.

[33] Zhang Jia’an, Li Fengxian, Wang Tiecheng, et al. A load prediction method using memory neural network and curve shape correction[J]. Electric Power Engineering Technology, 2024, 43(1):117-126(in Chinese).

[34] 王炜, 刘宏伟, 陈永杰, 等. 基于LSTM循环神经网络的风力发电预测[J]. 可再生能源, 2020, 38(9):1187-1191.

[35] Wang Wei, Liu Hongwei, Chen Yongjie, et al. Wind power forecast based on LSTM cyclic neural network[J]. Renewable Energy Resources, 2020, 38(9):1187-1191(in Chinese).

[36] 庞伟林, 关兆雄, 李沐栩. 一种混合LSTM—SAM的居民电力负荷预测模型[J]. 人工智能科学与工程, 2024(2):40-49.

[37] Pang Weilin, Guan Zhaoxiong, Li Muxu. A hybrid LSTM—SAM model for predicting residential load[J]. Artificial Intelligence Science and Engineering, 2024(2):40-49(in Chinese).

[38] 刘颂凯, 崔梓琪, 杨超, 等. 基于时空注意机制与LSTM的暂态电压稳定评估[J]. 电力系统及其自动化学报, 2024, 36(6):82-91.

[39] Liu Songkai, Cui Ziqi, Yang Chao, et al. Transient voltage stability assessment based on spatial—temporal attention mechanism and LSTM[J]. Proceeding of the CSU—EPSA, 2024, 36(6):82-91(in Chinese).

[40] 王增平, 赵兵, 纪维佳, 等. 基于GRU—NN模型的短期负荷预测方法[J]. 电力系统自动化, 2019, 43(5):53-58.

[41] Wang Zengping， Zhao Bing, Ji Weijia, et al. Short—term load forecasting method based on GRU—NN model[J]. Automation of Electric Power Systems, 2019, 43(5):53-58(in Chinese).

[42] Cybenko G. Approximation by superpositions of a sigmoidal function[J]. Mathematics of Control，Signals and Systems, 1989, 2:303-314.

基金资助

国家电网公司总部科技项目(5100-202314007A-1-1-ZN)