长短期记忆网络和生成对抗网络混合驱动的牵引供电系统关键运行数据增强方法

doi:10.19595/j.cnki.1000-6753.tces.250186

Abstract
Figure/Table
References
Related Citation (15)

Download: PDF (3020 KB) HTML (1 KB)
Export: BibTeX | EndNote (RIS)

Abstract In modern railway transportation systems, the efficient operation and precise control of traction power supply systems are critical for ensuring safety, reliability, and performance. However, existing data acquisition systems, which operate at second-level frequencies, are insufficient to meet the demands of modern traction power supply systems, particularly in scenarios involving high-speed trains, renewable energy integration, and complex fault conditions. This paper proposes an innovative data enhancement method based on a hybrid model that combines long short-term memory (LSTM) networks and generative adversarial networks (GANs).
The proposed method targets two critical issues: (1) the difficulty in capturing complex multivariate correlations among key operational parameters, (2) the inability of existing data enhancement methods to handle the long-term and short-term dependencies in time-series data effectively. Regarding the first challenge, the paper integrates principal component analysis (PCA) and grey relational analysis (GRA) to construct a measurement enhancement matrix. This matrix not only reduces data dimensionality but also thoroughly explores the correlations among variables, thereby enhancing the ability to analyze trends and dynamic changes in the data. For the second challenge, the paper introduces an LSTM-GAN model that leverages the temporal dependency- capturing capabilities of LSTM networks and the high-quality data generation capabilities of GANs. This hybrid model accurately captures the dynamic changes and deep dependency structures in measurement data, achieving high-precision time-series data enhancement. The LSTM-GAN model is designed to overcome the limitations of traditional GANs in handling time-series data. By integrating LSTM networks, the model effectively captures long-term dependencies to analyze transient phenomena such as overvoltage and short-circuit faults during bow-net disconnection. The model’s architecture includes an LSTM module that extracts temporal trends from the input data, which are then used as conditional inputs to the GAN. It allows the generator to produce synthetic data that aligns with the time-series context, ensuring high fidelity to the original data. Additionally, the model incorporates upper and lower bound constraints in the loss function to ensure generated data remains within realistic ranges, further enhancing its applicability to real-world scenarios.
The proposed method was validated using real-time data collected from the onboard monitoring system of a high-speed railway section in Northern China. The results demonstrate significant improvements in data accuracy and variable correlation analysis compared to other data enhancement algorithms. Specifically, the absolute error is reduced by an average of 0.019, and the data generation frequency is enhanced from seconds to milliseconds. The mean absolute error (MAE) and mean absolute percentage error (MAPE) for both current and voltage enhancement using the LSTM-GAN model were significantly lower than those of the other methods.

Key words： Traction power supply system data enhancement long short term memory-generative adversarial network (LSTM-GAN) model principal component analysis grey correlation analysis

Received: 27 January 2025

PACS:

TM922

	Service

	E-mail this article
	Add to my bookshelf
	Add to citation manager
	E-mail Alert
	RSS
	Articles by authors
	Ge Leijiao
	Lin Keyuan
	Ren Limiao

Cite this article:

Ge Leijiao,Lin Keyuan,Ren Limiao. Key Operation Data Enhancement Method of Traction Power Supply System Driven by Hybrid Long Short-Term Memory Network and Generative Adversarial Network[J]. Transactions of China Electrotechnical Society, 2026, 41(2): 607-621.

URL:

https://dgjsxb.ces-transaction.com/EN/10.19595/j.cnki.1000-6753.tces.250186 OR https://dgjsxb.ces-transaction.com/EN/Y2026/V41/I2/607

[1] He Xiaoqiong, Peng Jun, Han Pengcheng, et al.A novel advanced traction power supply system based on modular multilevel converter[J]. IEEE Access, 2019, 7: 165018-165028.
[2] 何晓琼, 韩鹏程, 王怡, 等. 基于级联-并联变换器的贯通式牵引变电所系统研究[J]. 铁道学报, 2017, 39(8): 52-61.
He Xiaoqiong, Han Pengcheng, Wang Yi, et al.Study on advanced cophase traction power substation system based on cascade-parallel converter[J]. Journal of the China Railway Society, 2017, 39(8): 52-61.
[3] 陈冲, 贾利民, 赵天宇, 等. 光伏和储能植入铁路牵引供电系统的拓扑架构与控制策略研究综述[J]. 电工技术学报, 2024, 39(24): 7874-7901.
Chen Chong, Jia Limin, Zhao Tianyu, et al.Research review on topology and control strategy of PV and energy storage connected to railway traction power supply systems[J]. Transactions of China Electro- technical Society, 2024, 39(24): 7874-7901.
[4] 陈艳波, 刘宇翔, 田昊欣, 等. 基于广义目标级联法的多牵引变电站光伏-储能协同规划配置[J]. 电工技术学报, 2024, 39(15): 4599-4612.
Chen Yanbo, Liu Yuxiang, Tian Haoxin, et al.Colla- borative planning and configuration of photovoltaic- energy storage in multi-traction substations based on generalized target cascading method[J]. Journal of Electrical Technology, 2024, 39(15): 4599-4612.
[5] 陈冲, 贾利民, 赵天宇, 等. 去碳化导向的轨道交通与新能源融合发展——形态模式、解决方案和使/赋能技术[J]. 电工技术学报, 2023, 38(12): 3321-3337.
Chen Chong, Jia Limin, Zhao Tianyu, et al.Decarburization-oriented rail transit and new energy integration development-morphological model, solution and enabling/enabling technology[J]. Journal of Elec- trical Technology, 2023, 38(12): 3321-3337.
[6] Li Bin, Chen Dongyang, He Jiawei, et al.Fault analysis and protection in flexible DC traction power supply system[J]. IEEE Transactions on Trans- portation Electrification, 2024, 10(3): 5599-5613.
[7] 肖嵩, 段君璋, 朱涛, 等. 高速铁路弓网离线过电压对车体电位的影响[J]. 中国铁道科学, 2023, 44(5): 180-190.
Xiao Song, Duan Junzhang, Zhu Tao, et al.Impact of pantograph-catenary off-line overvoltage on car body potential in high-speed railway[J]. China Railway Science, 2023, 44(5): 180-190.
[8] 肖嵩, 曹野, 吴广宁, 等. 高铁过电压对车载牵引供电系统的影响机理及抑制方法: 系统性综述[J]. 中国电机工程学报, 2024, 44(12): 4682-4702.
Xiao Song, Cao Ye, Wu Guangning, et al.Influence mechanism and suppression methodologies of the overvoltage of high-speed railway on the vehicle- mounted traction power supply system: systematic review[J]. Proceedings of the CSEE, 2024, 44(12): 4682-4702.
[9] 刘林青, 葛云龙, 李梦宇, 等. 基于量测数据和数据驱动技术的配电变压器状态监测与故障诊断[J]. 高压电器, 2020, 56(9): 11-19.
Liu Linqing, Ge Yunlong, Li Mengyu, et al.Condition monitoring and fault diagnosis of dis- tribution transformer based on measurement data and data-driven technology[J]. High Voltage Apparatus, 2020, 56(9): 11-19.
[10] 崔昊杨, 蔡杰, 陈磊, 等. 基于颜色编码的非侵入式负荷细粒度识别方法[J]. 电网技术, 2022, 46(4): 1557-1567.
Cui Haoyang, Cai Jie, Chen Lei, et al.Non-intrusive load fine-grained identification based on color encoding[J]. Power grid technology, 2022, 46(4): 1557-1567.
[11] 杨挺, 李大帅, 蔡绍堂, 等. 面向用户隐私保护的用电数据压缩加密方法[J]. 中国电机工程学报, 2022, 42(增刊1): 58-69.
Yang Ting, Li Dashuai, Cai Shaotang, et al.Non- intrusive load fine-grained identification based on color encoding[J]. Proceedings of the CSEE, 2022, 42(S1): 58-69.
[12] 马广富, 高升, 郭延宁. 一类伴有部分解耦干扰的非线性系统故障诊断[J]. 控制理论与应用, 2024, 41(2): 240-248.
Ma Guangfu, Gao Sheng, Guo Yanning.Fault diagnosis design for nonlinear systems corrupted by partially decoupled disturbances[J]. Control Theory & Applications, 2024, 41(2): 240-248.
[13] 谢庆, 张煊宇, 王春鑫, 等. 新一代人工智能技术在输变电设备状态评估中的应用现状及展望[J]. 高压电器, 2022, 58(11): 1-16.
Xie Qing, Zhang Xuanyu, Wang Chunxin, et al.Application status and prospect of the new generation artificial intelligence technology in the state evaluation of power transmission and transformation equipment[J]. High Voltage Apparatus, 2022, 58(11): 1-16.
[14] Gong Peng, Cao Yuan, Cai Baigen, et al.Multi- information location data fusion system of railway signal based on cloud computing[J]. Future Gen- eration Computer Systems, 2018, 88: 594-598.
[15] 刘洋, 尹彦宏, 杨斯泐, 等. 重载铁路牵引变电所数字孪生技术研究与应用[J]. 铁道运输与经济, 2023, 45(11): 106-114.
Liu Yang, Yin Yanhong, Yang Sile, et al.Research and application of digital twin technology in heavy-haul railwaytraction substations[J]. Railway Transportation and Economy, 2023, 45(11): 106-114.
[16] 王续卓, 李正烁, 邢家维, 等. 面向低感知度三相配电网的数据增强状态估计[J]. 中国电机工程学报, 2025, 45(15): 5942-5952.
Wang Xuzhuo, Li Zhengshuo, Xing Jiawei, et al.Data-augmented state estimation for partially visible three-phase distribution networks[J]. Proceedings of the CSEE, 2025, 45(15): 5942-5952.
[17] 朱超. 智能变电站网络采样中关键技术的研究[D]. 南京: 东南大学, 2014.
Zhu Chao.Research on key technologies in network sampling of intelligent substation[D]. Nanjing: Southeast University, 2014.
[18] Li Chen, Wang Kechong, Piao Yinchuan, et al.Surface micro-morphology model involved in grinding of GaN crystals driven by strain-rate and abrasive coupling effects[J]. International Journal of Machine Tools and Manufacture, 2024, 201: 104197.
[19] 李弈, 张金龙, 漆汉宏, 等. 基于变分深度嵌入-带有梯度惩罚的生成对抗网络的锂离子电池老化特性建模[J]. 电工技术学报, 2024, 39(13): 4226-4239.
Li Yi, Zhang Jinlong, Qi Hanhong, et al.Ageing performance modeling of Li-ion batteries based on variational deep embedding-Wasserstein GAN with gradient penalty[J]. Transactions of China Electro- technical Society, 2024, 39(13): 4226-4239.
[20] 朱玲, 李威, 王骞, 等. 基于校正条件生成对抗网络的风电场群绿氢储能系统容量配置[J]. 电工技术学报, 2024, 39(3): 714-730.
Zhu Ling, Li Wei, Wang Qian, et al.Wind farms- green hydrogen energy storage system capacity sizing method based on corrected-conditional generative adversarial network[J]. Transactions of China Elec- trotechnical Society, 2024, 39(3): 714-730.
[21] 李富盛, 林丹, 余涛, 等. 基于改进生成式对抗网络的电气数据升频重建方法[J]. 电力系统自动化, 2022, 46(3): 105-112.
Li Fusheng, Lin Dan, Yu Tao, et al.Frequency- increased reconstruction method for electrical data based on improved generative adversarial network[J]. Automation of Electric Power Systems, 2022, 46(3): 105-112.
[22] 杨玉莲, 齐林海, 王红, 等. 基于生成对抗和双重语义感知的配电网量测数据缺失重构[J]. 电力系统自动化, 2020, 44(18): 46-54.
Yang Yulian, Qi Linhai, Wang Hong, et al.Reconstruction of missing measurement data in distribution network based on generative adversarial network and double semantic perception[J]. Auto- mation of Electric Power Systems, 2020, 44(18): 46-54.
[23] Mi Jiaqi, Ma Congcong, Zheng Lihua, et al.WGAN- CL: a Wasserstein GAN with confidence loss for small-sample augmentation[J]. Expert Systems with Applications, 2023, 233: 120943.
[24] Ren Lei, Wang Haiteng, Laili Yuanjun.Diff-MTS: temporal-augmented conditional diffusion-based AIGC for industrial time series toward the large model era[J]. IEEE Transactions on Cybernetics, 2024, 54(12): 7187-7197.
[25] 王守相, 陈海文, 潘志新, 等. 采用改进生成式对抗网络的电力系统量测缺失数据重建方法[J]. 中国电机工程学报, 2019, 39(1): 56-64, 320.
Wang Shouxiang, Chen Haiwen, Pan Zhixin, et al.A reconstruction method for missing data in power system measurement using an improved generative adversarial network[J]. Proceedings of the CSEE, 2019, 39(1): 56-64, 320.
[26] Loey M, Manogaran G, Khalifa N E M. A deep transfer learning model with classical data augmentation and CGAN to detect COVID-19 from chest CT radiography digital images[J]. Neural Computing & Applications, 2020: 1-13.
[27] Ding Xin, Wang Yongwei, Xu Zuheng, et al.Distilling and transferring knowledge via cGAN- generated samples for image classification and regression[J]. Expert Systems with Applications, 2023, 213: 119060.
[28] Wang Pengyu, Zhu Hongqing, Huang Hui, et al.TMS- GAN: a twofold multi-scale generative adversarial network for single image dehazing[J]. IEEE Transa- ctions on Circuits and Systems for Video Technology, 2022, 32(5): 2760-2772.
[29] Liu Jie, Deng Wenfeng, Yang Chunhua, et al.SI- LSGAN: Complex network structure inference based on least square generative adversarial network[J]. Chaos, Solitons & Fractals, 2023, 173: 113739.
[30] Tov O, Alaluf Y, Nitzan Y, et al.Designing an encoder for StyleGAN image manipulation[J]. ACM Transactions on Graphics, 2021, 40(4): 1-14.
[31] Yang Ping, Li Shichao, Qin Shanyong, et al.Smart grid enterprise decision-making and economic benefit analysis based on LSTM-GAN and edge computing algorithm[J]. Alexandria Engineering Journal, 2024, 104: 314-327.
[32] 刘鑫蕊, 常鹏, 孙秋野. 基于XGBoost和无迹卡尔曼滤波自适应混合预测的电网虚假数据注入攻击检测[J]. 中国电机工程学报, 2021, 41(16): 5462-5476.
Liu Xinrui, Chang Peng, Sun Qiuye.Grid false data injection attacks detection based on XGBoost and unscented Kalman filter adaptive hybrid prediction[J]. Proceedings of the CSEE, 2021, 41(16): 5462-5476.
[33] 李大虎, 曹一家. 基于SCADA/PMU混合量测的广域动态实时状态估计方法[J]. 电网技术, 2007, 27(6): 72-78.
Li Dahu, Cao Jia.Wide-area real-time dynamic state estimation method based on hybrid SCADA/PMU measurements[J]. Power System Technology, 2007, 27(6): 72-78.
[34] 张润宝, 杨志鹏. 接触网运行状态检测监测系统研究与实践[J]. 中国铁路, 2019(9): 64-70.
Zhang Runbao, Yang Zhipeng.Research and practice of operation state inspection and monitoring system of overhead contact line system[J]. China Railway, 2019(9): 64-70.
[35] 王雁凌, 吴梦凯, 周子青, 等. 基于改进灰色关联度的电力负荷影响因素量化分析模型[J]. 电网技术, 2017, 41(6): 1772-1778.
Wang Yanling, Wu Mengkai, Zhou Ziqing, et al.Quantitative analysis model of power load influencing factors based on improved grey relational degree[J]. Power System Technology, 2017, 41(6): 1772-1778.
[36] Greff K, Srivastava R K, Koutník J, et al.LSTM: a search space odyssey[J]. IEEE Transactions on Neural Networks and Learning Systems, 2017, 28(10): 2222-2232.
[37] 刘伟, 王洪志. 基于改进注意力机制的时间卷积网络-长短期记忆网络短期电力负荷预测[J]. 电气技术, 2024, 25(10): 8-14.
Liu Wei, Wang Hongzhi.Short term power load forecasting based on temporal convolutional network- long short term memory and improved attention mechanism[J]. Electrical Engineering, 2024, 25(10): 8-14.
[38] Creswell A, White T, Dumoulin V, et al.Generative adversarial networks: an overview[J]. IEEE Signal Processing Magazine, 2018, 35(1): 53-65.