基于粒子群优化堆叠降噪自编码器的电力设备状态数据质量提升

doi:10.16183/j.cnki.jsjtu.2023.388

摘要/Abstract

摘要：

当下电力设备状态大数据呈现爆炸式增长,设备故障、数据传输以及人为操作失误等原因都会导致问题数据的出现,影响数据质量以及后续分析结果,因此数据清洗具有重要意义.目前大多数研究着力于识别异常数据并直接剔除,破坏了数据的完整性.针对此问题,提出一种基于改进堆叠降噪自编码器的数据清洗方法.首先,采用粒子群算法优化堆叠降噪自编码器中的超参数;然后,利用堆叠降噪自编码器提取、还原数据特征的特点来进行数据清洗,实现对孤立点的修复和对空缺数据的填补,以有效提升电力设备状态数据的质量.所提方法简单高效,可以同时提高数据集的准确性和完整性.以电力设备的历史运行数据为例进行测试,算例结果表明所提方法相比于其他经典方法,数据清洗效果更好,且针对不同异常程度和运行状态的数据集都有良好的清洗效果,能够提高电力设备状态数据的质量.

关键词: 电力设备, 状态数据, 堆叠降噪自编码器, 数据清洗

Abstract:

Big data related to power equipment condition is experiencing explosive growth. However, equipment failures and personnel errors result in dirty data, having a negative effect on data quality and subsequent analysis results. Therefore, data cleaning is of great significance. Most existing research focuses on direct identification and elimination of abnormal data, which compromises the integrity of the data. In order to solve this problem, a data cleaning method based on improved stack noise reduction autoencoder is proposed in this paper. First, particle swarm optimization is used to optimize the hyperparameters of the stack noise reduction autoencoder. Then, the characteristics of the autoencoder is used to extract and restore the data features to clean the data. The method improves data quality of power equipment condition by repairing isolated data points and filling in missing data, which is simple and efficient for improving the accuracy and integrity of the data set. Finally, the historical operation data of power equipment is taken as an example. The simulation results show that the proposed method outperforms other classical methods providing good cleaning results for data sets with different abnormal degrees in different running states. The proposed method offers an effective solution for improving the quality of power equipment status data effectively.

Key words: power equipment, status data, stacked denoising autoencoder, data cleaning

中图分类号:

TM377

计蓉, 侯慧娟, 盛戈皞, 张立静, 舒博, 江秀臣. 基于粒子群优化堆叠降噪自编码器的电力设备状态数据质量提升[J]. 上海交通大学学报, 2025, 59(6): 780-788.

JI Rong, HOU Huijuan, SHENG Gehao, ZHANG Lijing, SHU Bo, JIANG Xiuchen. Data Quality Improvement Method for Power Equipment Condition Based on Stacked Denoising Autoencoders Improved by Particle Swarm Optimization[J]. Journal of Shanghai Jiao Tong University, 2025, 59(6): 780-788.

图/表 10

图1

图2

图3

图4

图5

表1

表2

图6

表3

图7

参考文献 26

[1]	江秀臣, 盛戈皞. 电力设备状态大数据分析的研究和应用[J]. 高电压技术, 2018, 44(4): 1041-1050.
	JIANG Xiuchen, SHENG Gehao. Research and application of big data analysis of power equipment condition[J]. High Voltage Engineering, 2018, 44(4): 1041-1050.
[2]	陈敬德, 盛戈皞, 吴继健, 等. 大数据技术在智能电网中的应用现状及展望[J]. 高压电器, 2018, 54(1): 35-43.
	CHEN Jingde, SHENG Gehao, WU Jijian, et al. Application status and prospect of big data technology in smart grid[J]. High Voltage Apparatus, 2018, 54(1): 35-43.
[3]	方静, 彭小圣, 刘泰蔚, 等. 电力设备状态监测大数据发展综述[J]. 电力系统保护与控制, 2020, 48(23): 176-186.
	FANG Jing, PENG Xiaosheng, LIU Taiwei, et al. Development trend and application prospects of big data-based condition monitoring of power apparatus[J]. Power System Protection & Control, 2020, 48(23): 176-186.
[4]	李刚, 张博, 赵文清, 等. 电力设备状态评估中的数据科学问题: 挑战与展望[J]. 电力系统自动化, 2018, 42(21): 10-20.
	LI Gang, ZHANG Bo, ZHAO Wenqing, et al. Data science issues in state evaluation of power equipment: Challenges and prospects[J]. Automation of Electric Power Systems, 2018, 42(21): 10-20.
[5]	王军, 李建勋, 韩山, 等. 一种效能评估中缺失数据的填充方法[J]. 上海交通大学学报, 2017, 51(2): 180-185.
	WANG Jun, LI Jianxun, HAN Shan, et al. A missing data filling method for effectiveness evaluation system[J]. Journal of Shanghai Jiao Tong University, 2017, 51(2): 180-185.
[6]	周远翔, 林孟龙, 陈健宁, 等. 基于自注意力生成对抗网络的电力设备在线监测缺失数据填补[J]. 高电压技术, 2023, 49(5): 1795-1809.
	ZHOU Yuanxiang, LIN Menglong, CHEN Jianning, et al. Missing data imputation for online monitoring of power equipment based on self-attention generative adversarial networks[J]. High Voltage Engineering, 2023, 49(5): 1795-1809.
[7]	吴金娥, 王若愚, 段倩倩, 等. 基于反向k近邻过滤异常的群数据异常检测[J]. 上海交通大学学报, 2021, 55(5): 598-606. doi: 10.16183/j.cnki.jsjtu.2020.011
	WU Jin’e, WANG Ruoyu, DUAN Qianqian, et al. Collective data anomaly detection based on reverse k-nearest neighbor filtering[J]. Journal of Shanghai Jiao Tong University, 2021, 55(5): 598-606.
[8]	李富盛, 陈伟松, 钱斌, 等. 面向低压配电网智能电表误差监测的LightGBM-EM-EC多变量缺失数据高效重建[J]. 中国电机工程学报, 2022, (42 Sup.1): 95-105.
	LI Fusheng, CHEN Weisong, QIAN Bin, et al. Efficient reconstruction of multivariate missing data for smart meter error monitoring in low voltage distribution network based on LightGBM-EM-EC[J]. Proceedings of the CSEE. 2022, (Sup.1): 95-105.
[9]	杨茂, 张书天, 王天硕, 等. 基于IKLIEP-四分位模型的风电场异常数据识别算法[J]. 高电压技术, 2023, 49(7): 2952-2960.
	YANG Mao, ZHANG Shutian, WANG Tianshuo, et al. Identification algorithm of wind farm abnormal data based on IKLIEP-quartile model[J]. High Voltage Engineering, 2023, 49(7): 2952-2960.
[10]	王文森, 杨晓西, 刘阳, 等. 基于层次聚类分析的变压器油中溶解气体在线监测数据异常检测[J]. 高压电器, 2023, 59(1): 142-147.
	WANG Wensen, YANG Xiaoxi, LIU Yang, et al. Anomaly detection of online monitoring data of dissolved gases in transformer oil based on hierarchical cluster analysis[J]. High Voltage Apparatus, 2023, 59(1): 142-147.
[11]	杨挺, 孙兆帅, 季浩, 等. 基于矩阵范数优化理论的用电数据质量提升算法[J]. 中国电机工程学报, 2022, 42(10): 3501-3511.
	YANG Ting, SUN Zhaoshuai, JI Hao, et al. Electricity consumption data quality improvement algorithm based on matrix norm optimization theory[J]. Proceedings of the CSEE, 2022, 42(10): 3501-3511.
[12]	娄建楼, 胥佳, 陆恒, 等. 基于功率曲线的风电机组数据清洗算法[J]. 电力系统自动化, 2016, 40(10): 116-121.
	LOU Jianlou, XU Jia, LU Heng, et al. Wind turbine data-cleaning algorithm based on power curve[J]. Automation of Electric Power Systems, 2016, 40(10): 116-121.
[13]	林峻, 严英杰, 盛戈皞, 等. 考虑时间序列关联的变压器在线监测数据清洗[J]. 电网技术, 2017, 41(11): 3733-3740.
	LIN Jun, YAN Yingjie, SHENG Gehao, et al. Online monitoring data cleaning of transformer considering time series correlation[J]. Power System Technology, 2017, 41(11): 3733-3740.
[14]	朱有产, 梁玮轩, 王英姿. 基于时间序列的配电网数据清理和融合方法研究[J]. 电网技术, 2021, 45(7): 2839-2846.
	ZHU Youchan, LIANG Weixuan, WANG Yingzi. Research on data cleaning and fusion in distribution power grid based on time series technology[J]. Power System Technology, 2021, 45(7): 2839-2846.
[15]	REN C, XU Y. A fully data-driven method based on generative adversarial networks for power system dynamic security assessment with missing data[J]. IEEE Transactions on Power Systems, 2019, 34(6): 5044-5052.
[16]	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.
	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.
[17]	代杰杰, 宋辉, 杨祎, 等. 基于栈式降噪自编码器的输变电设备状态数据清洗方法[J]. 电力系统自动化, 2017, 41(12): 224-230.
	DAI Jiejie, SONG Hui, YANG Yi, et al. Cleaning method for status data of power transmission and transformation equipment based on stacked denoising autoencoders[J]. Automation of Electric Power Systems, 2017, 41(12): 224-230.
[18]	SHIN H C, ORTON M R, COLLINS D J, et al. Stacked autoencoders for unsupervised feature learning and multiple organ detection in a pilot study using 4D patient data[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2013, 35(8): 1930-1943.
[19]	HONG C Q, YU J, WAN J, et al. Multimodal deep autoencoder for human pose recovery[J]. IEEE Transactions on Image Processing: A publication of the IEEE Signal Processing Society, 2015, 24(12): 5659-5670.
[20]	VINCENT P, LAROCHELLE H, BENGIO Y, et al. Extracting and composing robust features with denoising autoencoders[C]// Proceedings of the 25th international conference on Machine learning. Helsinki, Finland: ACM, 2008: 1096-1103.
[21]	VINCENT P, LAROCHELLE H, LAJOIE I, et al. Stacked denoising autoencoders: Learning useful representations in a deep network with a local denoising criterion[J]. Journal of Machine Learning Research, 2010, 11: 3371-3408.
[22]	李兵, 梁舒奇, 单万宁, 等. 基于改进正余弦算法优化堆叠降噪自动编码器的电机轴承故障诊断[J]. 电工技术学报, 2022, 37(16): 4084-4093.
	LI Bing, LIANG Shuqi, SHAN Wanning, et al. Motor bearing fault diagnosis based on improved sine and cosine algorithm for stacked denoising autoencoders[J]. Transactions of China Electrotechnical Society, 2022, 37(16): 4084-4093.
[23]	招景明, 唐捷, 潘峰, 等. 基于SDAE和双模型联合训练的低压用户窃电检测方法[J]. 电测与仪表, 2021, 58(12): 161-168.
	ZHAO Jingming, TANG Jie, PAN Feng, et al. Detection method of electricity theft for low-voltage users based on SDAE and double-model joint training[J]. Electrical Measurement & Instrumentation, 2021, 58(12): 161-168.
[24]	KENNEDY J, EBERHART R. Particle swarm optimization[C]// Proceedings of ICNN’95-International Conference on Neural Networks, Perth, Australia: IEEE, 1995: 1942-1948.
[25]	王立文, 刘建华, 陈斌, 等. 基于改进的PSO优化SVR的机场道面积冰预测[J]. 计算机仿真, 2021, 38(9): 45-50.
	WANG Liwen, LIU Jianhua, CHEN Bin, et al. Prediction of airport road surface icing based on improved PSO optimized SVR[J]. Computer Simulation, 2021, 38(9): 45-50.
[26]	DU B G, HUANG S, GUO J, et al. Interval forecasting for urban water demand using PSO optimized KDE distribution and LSTM neural networks[J]. Applied Soft Computing, 2022, 122: 108875.

时间点	温度/℃					e_error/%
时间点	原始值	BP	PSO-SVM	SDAE	PSO-SDAE	e_error/%
6	52.700	45.702	48.431	54.043	53.265	1.07
17	52.512	45.702	58.470	53.964	53.813	2.48
24	56.676	61.560	60.217	59.378	59.362	4.74
33	58.665	61.560	60.736	59.378	59.365	1.19
41	54.534	45.702	48.194	53.963	53.813	1.32
42	53.253	45.702	48.193	53.963	53.813	1.05
54	53.160	45.702	57.311	53.963	53.235	0.14
66	58.421	61.553	60.602	53.963	59.353	1.60
80	49.070	61.560	58.633	59.378	51.362	4.67
90	56.073	45.702	58.469	53.963	53.813	4.03

方法	e_RMSE	e_MAE	e_MPAE/%
BP	2.4320	0.718	1.341
PSO-SVM	1.6110	0.613	1.140
SDAE	1.2100	0.353	0.666
PSO-SDAE	0.5148	0.304	0.560

方法	异常率5%		异常率10%		异常率20%		异常率30%
方法	e_RMSE	e_MPAE/%	e_RMSE	e_MPAE/%	e_RMSE	e_MPAE/%	e_RMSE	e_MPAE/%
BP	51.232	1.6295	72.677	3.5770	107.891	7.096	140.287	11.147
PSO-SVM	42.045	1.3640	70.409	3.6560	96.124	6.135	125.314	7.762
SDAE	38.429	3.3078	54.596	5.2928	80.977	7.043	73.823	5.231
PSO-SDAE	35.731	1.2963	53.246	2.7180	77.272	4.679	72.181	4.806