能源数字对象压缩流转变时间尺度异常检测方法

Variable Time-Scale Anomaly Detection Method for Energy Digital Object Compression and Flow

摘要

针对能源数字对象在供用能主体间流转效率低、异常检测实时性和准确度不高的问题,提出了一种基于压缩感知的数据流转和异常检测算法. 首先,基于压缩感知理论实现了能源数字对象数据压缩,并采用基于动态基追踪和最优稀疏度估计的广义正交匹配追踪(GOMP-DBP-OSE)算法进行数据重构,提高了数据在不同主体间的流转速度；其次,提出了多时间尺度下双向长短期记忆-Transformer(BiLSTM-Transformer)融合模型的数据异常滑动检测算法,同时解析数据流的长相关性和短相似性,提升了异常检测的实时性和准确度；最后,以某地区电动汽车-充电桩-电网互动的真实数据作为能源数字对象实验实例,验证了本文所提方法的有效性. 实验结果表明,所提压缩算法的压缩率相较于对比算法平均提升14.2%,重构速度相比传统变步长稀疏度自估计子空间追踪(VSSESP)和压缩采样匹配追踪(CoSaMP)算法分别提升了32.8%和44.5%,所提多时间尺度融合模型的异常检测准确率超过95%,相较于其他异常检测算法性能优越.

Abstract

To improve the flow efficiency of energy digital objects between energy suppliers and users and enhance low real-time performance and accuracy of anomaly detection, a data flow and anomaly detection algorithm based on compressed sensing is proposed. First, the data compression of energy digital objects is realized according to the theory of compressed sensing, after which the data reconstruction is carried out using the generalized orthogonal matching pursuit based on dynamic basis pursuit and optimal sparsity estimation(GOMP-DBP-OSE) algorithm, thus improving the speed of data flow between different subjects. Second, a data anomaly sliding detection algorithm is proposed for the bidirectional long short-term memory Transformer(BiLSTM-Transformer) fusion model under multi-time scales, thereby resolving the long correlation and short similarity of the data flow at the same time. This solution improves the accuracy of real-time anomaly detection. Finally, the real data of electric vehicle-charging pile-grid interaction in a region are used as an experimental example of an energy digital object to verify the effectiveness of the method proposed in this paper. The experimental results revealed that the compression rate of the proposed algorithm is on average 14.2% higher than that of the comparison algorithms, whereas the reconstruction speeds increase by 32.8% and 44.5% compared with those of the traditional variable step sparsity estimation subspace pursuit (VSSESP) and compressive sampling matching pursuit(CoSaMP) algorithms, respectively. Moreover, the accuracy of anomaly detection of the proposed multi-time scale fusion model exceeds 95%, which is superior to the performance of other anomaly detection algorithms.

关键词

能源数字对象 / 压缩感知 / 异常检测 / 双向长短期记忆

Key words

energy digital object(DO) / compressed sensing / anomaly detection / bidirectional long short-term memory(BiLSTM)

参考文献

[1] 余贻鑫. 面向21世纪的智能电网[J]. 天津大学学报(自然科学与工程技术版), 2020, 53(6):551-556.

[2] Yu Yixin. A brief description of the basics of the smart grid[J]. Journal of Tianjin University(Science and Technology), 2020, 53(6):551-556(in Chinese).

[3] 苏万华, 邱俊松, 邬斌扬, 等. 基于运行特性和生命周期理论的综合能源系统优化方法[J]. 天津大学学报(自然科学与工程技术版), 2022, 55(4):371-382.

[4] Su Wanhua, Qiu Junsong, Wu Binyang, et al. Integrated energy system optimization method based on the equipment operating characteristics and life cycle theory[J]. Journal of Tianjin University(Science and Technology), 2022, 55(4):371-382(in Chinese).

[5] 杨 挺, 刘亚闯, 刘宇哲, 等. 信息物理系统技术现状分析与趋势综述[J]. 电子与信息学报, 2021, 43(12):3393-3406.

[6] Yang Ting, Liu Yachuang, Liu Yuzhe, et al. Review on cyber—physical system：Technology analysis and trends[J]. Journal of Electronics & Information Technology, 2021, 43(12):3393-3406(in Chinese).

[7] Sharp C. Overview of the digital object architecture(DOA)[EB/OL]. https://www.internetsociety.org/resources/doc/2016/overview—of—the—digital—object—architecture—doa/, 2016—11—26.

[8] Sayao LF, Sales LF, Segundo W. Fair digital object：Expanding the boundaries of data stewardship[J]. Informacao & Sociedade—Estudos, 2023, 33:1-27.

[9] Reilly S, Tupelo—Schneck R. Digital object repository server：A component of the digital object architecture[J/OL]. D—Lib Magzine, https://www.dlib.org/dlib/january10/reilly/01reilly.html, 2010—01—02.

[10] Hisseine M A, Chen D J, Xiao Y, et al. A review of digital object architecture and handle system：Development,current applications and prospective[J]. Internet of Things, 2024, 26:101230.

[11] 赵少东, 王程斯. 基于多级索引集群均衡的电力能源大数据采集方法研究[J]. 电网与清洁能源, 2024, 40(8):85-90.

[12] Zhao Shaodong, Wang Chengsi. Research on power and energy big data acquisition methods based on multi—level index cluster equilibrium[J]. Power System and Clean Energy, 2024, 40(8):85-90(in Chinese).

[13] 王 伟, 周少泽, 黄 萌, 等. 构网型技术：演进历程、功能定位与应用展望[J]. 电力系统自动化, 2025, 49(1):1-13.

[14] Wang Wei, Zhou Shaoze, Huang Meng, et al. Grid—forming technologies：Evolution history,function,and application prospects[J]. Automation of Electric Power Systems, 2025, 49(1):1-13(in Chinese).

[15] 曾党泉. 基于PDC编码的中文文本压缩算法[J]. 计算机工程与应用, 2015, 51(17):205—209,227.

[16] Zeng Dangquan. Chinese text compression algorithm based on PDC coding[J]. Computer Engineering and Applications, 2015, 51(17):205—209,227(in Chinese).

[17] 华 强. 中文文本压缩的LZSSCH算法[J]. 中文信息学报, 1998, 12(1):50-56.

[18] Hua Qiang. The LZSSCH data compression algorithm for Chinese text files[J]. Journal of Chinese Information Processing, 1998, 12(1):50-56(in Chinese).

[19] 朱巧明, 赵英英, 钱培德. 基于中文词编码的压缩算法ZHCP的实现[J]. 小型微型计算机系统, 2003, 24(2):306-308.

[20] Zhu Qiaoming, Zhao Yingying, Qian Peide. To implement a compress algorithm ZHCP based upon Chinese word coding[J]. Mini—Micro Systems, 2003, 24(2):306-308(in Chinese).

[21] 杨 挺, 尚 昆, 袁 博, 等. 基于压缩感知的盲源信号分离检测方法[J]. 天津大学学报(自然科学与工程技术版), 2016, 49(11):1138-1143.

[22] Yang Ting, Shang Kun, Yuan Bo, et al. Blind signal separation detection method based on compressed sensing[J]. Journal of Tianjin University(Science and Technology), 2016, 49(11):1138-1143(in Chinese).

[23] 田金鹏, 刘小娟, 郑国莘. 一种变步长稀疏度自适应子空间追踪算法[J]. 自动化学报, 2016, 42(10):1512-1519.

[24] Tian Jinpeng, Liu Xiaojuan, Zheng Guoshen. A variable step size sparsity adaptive subspace pursuit algorithm[J]. Acta Automatica Sinica, 2016, 42(10):1512-1519(in Chinese).

[25] Yang Y Y, Zhu W. Research based on improved CNN—SVM fault diagnosis of V2G charging pile[J]. Electronics, 2023, 12(3):655.

[26] 邓 丰, 史鸿飞, 冯思旭, 等. CNN—LSTM 全景故障特征挖掘的配电网单端定位方法[J]. 中国电机工程学报, 2023, 43(增 1):114-126.

[27] Deng Feng, Shi Hongfei, Feng Sixu, et al. Single—ended traveling wave location method for distribution network based on CNN—LSTM panoramic fault feature mining[J]. Proceedings of the CSEE, 2023, 43(Suppl 1):114-126(in Chinese).

[28] 金 萍, 侯 娟. 面向新型电力系统的粗糙集和双流网络自动化物联设备故障诊断方法研究[J]. 电测与仪表, 2024, 61(9):166-171.

[29] Jin Ping, Hou Juan. Research on fault diagnosis method of rough set and dual—stream network automation IoT device for novel power system[J]. Electrical Measurement and Instrumentation, 2024, 61(9):166-171(in Chinese).

[30] 王卓瑜, 王 磊, 陆 婷, 等. 基于逆向运算法的海量大数据安全存储方法[J]. 兵工自动化, 2024, 43(7):23-26.

[31] Wang Zhuoyu, Wang Lei, Lu Ting, et al. Secure storage method of massive big data based on reverse operation[J]. Ordnance Automation, 2024, 43(7):23-26(in Chinese).

[32] 马 涛, 王仁智, 陈 丰, 等. 车—路协同下路面状态感知技术发展现状与展望[J]. 山东大学学报(工学版), 2024, 54(5):1-28.

[33] Ma Tao, Wang Renzhi, Chen Feng, et al. Development status and prospects of pavement state sensing technology under vehicle—road interaction[J]. Journal of Shandong University(Engineering Edition), 2024, 54(5):1-28(in Chinese).

[34] 杨 挺, 杨风霞, 叶芷杉, 等. 基于动态采样压缩感知的超谐波监测方法[J]. 中国电机工程学报, 2023, 43(16):6278-6288.

[35] Yang Ting, Yang Fengxia, Ye Zhishan, et al. A new measurment method for supra harmonics based on dynamic sampling and compressed sensing[J]. Proceedings of the CSEE, 2023, 43(16):6278-6288(in Chinese).

[36] 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 1):58-69.

[37] Yang Ting, Li Dashuai, Cai Shaotang, et al. Power data compression and encryption method for user privacy protection[J]. Proceedings of the CSEE, 2022, 42(Suppl 1):58-69(in Chinese).

[38] Wang J, Kwon S, Shim B. Generalized orthogonal matching pursuit[J]. IEEE Transactions on Signal Processing, 2012, 60(12):6202-6216.

基金资助

国家电网有限公司总部科技资助项目(5700-202390591A-3-2-ZN)