基于特征分析与MGAT-TCN模型的非侵入式户用光伏辨识

倪旖旎; 邬峥涛; 李子晨; 夏杨红

doi:10.19912/j.0254-0096.tynxb.2024-2387

PDF(3375 KB)

太阳能学报 ›› 2026, Vol. 47 ›› Issue (5) : 683-696. DOI: 10.19912/j.0254-0096.tynxb.2024-2387

基于特征分析与MGAT-TCN模型的非侵入式户用光伏辨识

倪旖旎¹, 邬峥涛², 李子晨¹, 夏杨红¹

作者信息 +

NON-INTRUSIVE RESIDENTIAL PHOTOVOLTAIC IDENTIFICATION BASED ON FEATURE ANALYSIS AND MGAT-TCN MODEL

Ni Yini¹, Wu Zhengtao², Li Zichen¹, Xia Yanghong¹

Author information +

文章历史 +

摘要

该文针对未加装单独监测装置的表后小型分布式光伏设备,以电表净负荷数据为基础,提出一种基于特征分析和多头图注意力网络和时间卷积网络（MGAT-TCN）模型的非侵入式户用光伏辨识方法。首先,在对光伏设备并网及状态切换过程仿真分析的基础上,提出基于多功能复合滑动窗的两阶段CUSUM事件检测算法对光伏状态切换与并网行为进行实时定位;其次,提出基于MGAT-TCN光伏辨识模型,该方法能通过多头图注意力网络有效处理光伏设备相互耦合的多维特征,从而提高辨识准确率与效率。最后,方法于自搭建非侵入式实验平台进行测试。结果表明,所提事件检测算法能够准确定位光伏行为,且相较于传统事件检测算法减少了至少0.3 s的时间定位误差;同时,所提的MGAT-TCN辨识模型对光伏设备的辨识准确率高于典型的神经网络方法至少2.66个百分点,且可达到96%以上。

Abstract

This paper focuses on behind-the-meter PV monitoring and proposes a non-intrusive residential PV identification method based on feature analysis and a hybrid model combining a Multi-Graph Attention Network with a Temporal Convolutional Network （MGAT-TCN）. First, the grid-connection and state-switching characteristics of household two-stage PV devices are analyzed to construct and rank a comprehensive feature set. Next, a two-stage cumulative sum （CUSUM） event detection algorithm based on a multifunctional composite sliding window, is proposed to detect PV state-switching and grid-connecting behaviors. Subsequently, the MGAT-TCN model is employed to accurately identify these behaviors using an optimized subset of features. Finally, the accuracy and real-time performance of the proposed method are validated on a self-built non-intrusive experimental platform. The results demonstrate that the proposed event detection algorithm accurately locates PV behaviors, reducing time-localization errors by at least 0.3s compared to traditional algorithms. Furthermore, the MGAT-TCN model achieves an identification accuracy of over 96%, outperforming typical neural network methods by at least 2.66%.

导出引用

倪旖旎, 邬峥涛, 李子晨, 夏杨红. 基于特征分析与MGAT-TCN模型的非侵入式户用光伏辨识[J]. 太阳能学报. 2026, 47(5): 683-696 https://doi.org/10.19912/j.0254-0096.tynxb.2024-2387

Ni Yini, Wu Zhengtao, Li Zichen, Xia Yanghong. NON-INTRUSIVE RESIDENTIAL PHOTOVOLTAIC IDENTIFICATION BASED ON FEATURE ANALYSIS AND MGAT-TCN MODEL[J]. Acta Energiae Solaris Sinica. 2026, 47(5): 683-696 https://doi.org/10.19912/j.0254-0096.tynxb.2024-2387

中图分类号： TM714

参考文献

[1] 陈国平, 董昱, 梁志峰. 能源转型中的中国特色新能源高质量发展分析与思考[J]. 中国电机工程学报, 2020, 40(17): 5493-5505.
CHEN G P, DONG Y, LIANG Z F.Analysis and reflection on high-quality development of new energy with Chinese characteristics in energy transition[J]. Proceedings of the CSEE, 2020, 40(17): 5493-5505.
[2] 国家能源局. 2025年可再生能源并网运行情况[EB/OL]. https://www.nea.gov.cn/20260212/742b8c6a078347b0b39de676c05c5d58/c.html.
[3] 李军徽, 孙大朋, 朱星旭, 等. 光伏高渗透率下分布式储能群间协同的电压控制策略[J]. 电力系统自动化, 2023, 47(10): 47-56.
LI J H, SUN D P, ZHU X X, et al.Voltage regulation strategy for distributed energy storage considering coordination among clusters with high penetration of photovoltaics[J]. Automation of electric power systems, 2023, 47(10): 47-56.
[4] 栗峰, 丁杰, 周才期, 等. 新型电力系统下分布式光伏规模化并网运行关键技术探讨[J]. 电网技术, 2024, 48(1): 184-196.
LI F, DING J, ZHOU C Q, et al.Key technologies of large-scale grid-connected operation of distributed photovoltaic under new-type power system[J]. Power system technology, 2024, 48(1): 184-196.
[5] 冯靖雷, 李正烁. 表后光伏发电辨识方法综述[J]. 电网技术, 2025, 49(5): 1983-1992.
FENG J L, LI Z S.Review of behind-the-meter photovoltaic generation identification methods[J]. Power system technology, 2025, 49(5): 1983-1992.
[6] HART G W.Nonintrusive appliance load monitoring[J]. Proceedings of the IEEE, 1992, 80(12): 1870-1891.
[7] TRUONG C, OUDRE L, VAYATIS N.Selective review of offline change point detection methods[J]. Signal processing, 2020, 167: 107299.
[8] 汪颖, 杨维, 肖先勇, 等. 基于U-I轨迹曲线精细化识别的非侵入式居民负荷监测方法[J]. 电网技术, 2021, 45(10): 4104-4113.
WANG Y, YANG W, XIAO X Y, et al.Non-intrusive residential load monitoring method based on refined identification of U-I trajectory curve[J]. Power system technology, 2021, 45(10): 4104-4113.
[9] JIANG C, SALLES D, XU W, et al.Assessing the collective harmonic impact of modern residential loads: part II: applications[J]. IEEE transactions on power delivery, 2012, 27(4): 1947-1955.
[10] LEE D.Phase noise as power characteristic of individual appliance for non-intrusive load monitoring[J]. Electronics letters, 2018, 54(16): 993-995.
[11] 周东国, 张恒, 周洪, 等. 基于状态特征聚类的非侵入式负荷事件检测方法[J]. 电工技术学报, 2020, 35(21): 4565-4575.
ZHOU D G, ZHANG H, ZHOU H, et al.Non-intrusive load event detection method based on state feature clustering[J]. Transactions of China Electrotechnical Society, 2020, 35(21): 4565-4575.
[12] 牟魁翌, 杨洪耕. 基于PLA-GDTW支持向量机的非侵入式负荷监测方法[J]. 电网技术, 2019, 43(11): 4185-4193.
MOU K Y, YANG H G.Non-intrusive load identification method based on PLA-GDTW support vector machine[J]. Power system technology, 2019, 43(11): 4185-4193.
[13] 于超, 覃智君, 阳育德. 基于启停状态识别改进因子隐马尔可夫模型的非侵入式负荷分解[J]. 电网技术, 2021, 45(11): 4540-4550.
YU C, QIN Z J, YANG Y D.Non-intrusive load disaggregation with improved factorial hidden Markov model considering ON-OFF status recognition[J]. Power system technology, 2021, 45(11): 4540-4550.
[14] 林顺富, 詹银枫, 李毅, 等. 基于CNN-BiLSTM与DTW的非侵入式住宅负荷监测方法[J]. 电网技术, 2022, 46(5): 1973-1981.
LIN S F, ZHAN Y F, LI Y, et al.Non-intrusive residential load monitoring method based on CNN-BiLSTM and DTW[J]. Power system technology, 2022, 46(5): 1973-1981.
[15] ENRÍQUEZ R, JIMÉNEZ M J, HERAS M R. Towards non-intrusive thermal load Monitoring of buildings: BES calibration[J]. Applied energy, 2017, 191: 44-54.
[16] 武昕, 于金莹, 彭林, 等. 基于用户边缘侧事件解析的工业电力负荷非侵入式感知辨识[J]. 电力系统自动化, 2021, 45(4): 29-37.
WU X, YU J Y, PENG L, et al.Non-intrusive perception and identification of industrial power load based on analysis of event on user edge[J]. Automation of electric power systems, 2021, 45(4): 29-37.
[17] XIANG Y, LIU J Y, LI R, et al.Economic planning of electric vehicle charging stations considering traffic constraints and load profile templates[J]. Applied energy, 2016, 178: 647-659.
[18] 施雨松, 徐青山, 郑建. 基于特征选择与增量学习的非侵入式电动自行车充电辨识方法[J]. 电力系统自动化, 2021, 45(7): 87-94.
SHI Y S, XU Q S, ZHENG J.Non-intrusive charging identification method for electric bicycles based on feature selection and incremental learning[J]. Automation of electric power systems, 2021, 45(7): 87-94.
[19] DINESH C, WELIKALA S, LIYANAGE Y, et al.Non-intrusive load monitoring under residential solar power influx[J]. Applied energy, 2017, 205: 1068-1080.
[20] MORENO JARAMILLO A F, LAVERTY D M, DEL RINCÓN J M, et al. Non-intrusive load monitoring algorithm for PV identification in the residential sector[C]//2020 31st Irish Signals and Systems Conference (ISSC). Letterkenny, Ireland, 2020: 1-6.
[21] ZHANG Y S, QIAN W Y, YE Y J, et al.A novel non-intrusive load monitoring method based on ResNet-seq2seq networks for energy disaggregation of distributed energy resources integrated with residential houses[J]. Applied energy, 2023, 349: 121703.
[22] WANG S Y, LI R, EVANS A, et al.Regional nonintrusive load monitoring for low voltage substations and distributed energy resources[J]. Applied energy, 2020, 260: 114225.
[23] 孙秋野, 刘月, 胡旌伟, 等. 基于GAN的非侵入式自能源建模[J]. 中国电机工程学报, 2020, 40(21): 6784-6793.
SUN Q Y, LIU Y, HU J W, et al.Non-intrusive we-energy modeling based on GAN technology[J]. Proceedings of the CSEE, 2020, 40(21): 6784-6793.
[24] 邵文权, 王猛, 吴朝俊, 等. 基于改进滑模控制的光伏系统MPPT控制策略[J]. 太阳能学报, 2021, 42(10): 87-93.
SHAO W Q, WANG M, WU C J, et al.MPPT control strategy based on improved sliding mode control for photovoltaic power system[J]. Acta energiae solaris sinica, 2021, 42(10): 87-93.
[25] 谢路耀, 莫晨飞, 陈怡. 基于并联光伏组件外特性的最大功率点跟踪方法[J]. 太阳能学报, 2022, 43(4): 148-153.
XIE L Y, MO C F, CHEN Y.MPPT method based on external characteristics of paralleled PV modules[J]. Acta energiae solaris sinica, 2022, 43(4): 148-153.
[26] 谢宁, 罗安, 马伏军, 等. 大型光伏电站与电网谐波交互影响[J]. 中国电机工程学报, 2013, 33(34): 9-16.
XIE N, LUO A, MA F J, et al.Harmonic interaction between large-scale photovoltaic power stations and grid[J]. Proceedings of the CSEE, 2013, 33(34): 9-16.
[27] 刘庆强. 居民负荷特性研究及应用[D]. 上海: 上海电力学院, 2014.
LIU Q Q.Research and application of residents'load characteristics[D]. Shanghai: Shanghai University of Electric Power, 2014.
[28] RESHEF D N, RESHEF Y A, FINUCANE H K, et al.Detecting novel associations in large data sets[J]. Science, 2011, 334(6062): 1518-1524.
[29] 史帅彬, 张恒, 邓世聪, 等. 基于复合滑动窗的CUSUM暂态事件检测算法[J]. 电测与仪表, 2019, 56(17): 13-18.
SHI S B, ZHANG H, DENG S C, et al.CUSUM transient event detection algorithm based on compound sliding window[J]. Electrical measurement & instrumentation, 2019, 56(17): 13-18.
[30] BAI S J, KOLTER J Z, KOLTUN V. An empirical evaluation of generic convolutional and recurrent networks for sequence modeling[EB/OL].2018: arXiv: 1803.01271. https://arxiv.org/abs/1803.01271.
[31] VELIKOVI P, CUCURULL G, CASANOVA A, et al.Graph attention networks[C]//International Conference on Learning Representations (ICLR). Vancouver, Canada, 2018: 1-12.
[32] WANG Y, HU L, WU Y, et al.Graph multihead attention pooling with self-supervised learning[J]. Entropy, 2022, 24(12): 1745.
[33] 赵洪山, 吴雨晨, 潘思潮, 等. 基于TCN和高斯过程残差建模学习的净负荷概率预测方法[J]. 太阳能学报, 2024, 45(12): 588-595.
ZHAO H S, WU Y C, PAN S C, et al.Probabilistic net load forecasting based on TCN and Gaussian process-enabled residual modeling learning approach[J]. Acta energiae solaris sinica, 2024, 45(12): 588-595.