作者：曹昆, 李国昌, 王艳松, 胡彩娥, 马龙飞

作者单位：国网北京市电力公司，北京 100075

关键词：大数据分析;输电设备;关键状态;高维随机矩阵;单环定律

摘要：

近年来，智能变电站不断兴起，对大量的数据处理要求也不断提高。为此，该文建立一种基于大数据技术和随机矩阵理论的新的变电站智能监测方法，通过提取变电站的关键状态和各参数之间的统计相关性，检测其运行状态。该方法主要通过时序高维随机矩阵进行推导，利用单环定律的半径作为统计分析表征测量数据，通过对这些统计分析的比较，完成关键状态的评估和异常检测。文章通过基于变电站自身变压器数据以及基于与变电站相连的输电线路各项参数进行建模分析，由此说明利用高维矩阵进行异常检测的有效性和可行性，与传统的阈值比较方法相比，所提方法采用高维矩阵的方法结合了状态变量时空间的关系，挖掘了数据的潜在发展趋势，具有较高的精度，在实际工程应用中具有较高的可参考性。

Substation state evaluation based on big data technology and random matrix theory
CAO Kun, LI Guochang, WANG Yansong, HU Caie, MA Longfei
State Grid Beijing Electric Power Company, Beijing 100075, China
Abstract: In recent years, intelligent substation is rising, and the requirements for a large number of data processing are also increasing. For this reason, a new intelligent monitoring method of substation based on big data technology and random matrix theory is established in this paper. By extracting the statistical correlation between key states and parameters of substation, the operation state of substation is detected. In this method, the high-dimensional random matrix of time series is used to conduct deduction, and the radius of single ring law is used as the statistical analysis to characterize the measurement data. Through the comparison of these statistical analysis, the evaluation of key states and anomaly detection are completed. Finally, through the modeling analysis based on the transformer data of the substation itself and the parameters of the transmission line connected with the substation, this paper shows the effectiveness and feasibility of anomaly detection using high-dimensional matrix. Compared with the traditional threshold comparison method, the proposed method uses high-dimensional matrix method combined with the time-space relationship of state variables to mine the data Potential development trend, with high precision, has a high reference in practical engineering applications.
Keywords: big data analysis;transmission equipment;key state;high-dimensional random matrix;single loop law
2020, 46(8):149-157  收稿日期: 2020-04-01;收到修改稿日期: 2020-05-08
基金项目:
作者简介: 曹昆（1976-），女，辽宁沈阳市人，高级工程师，硕士，主要研究方向为电力系统及其自动化
参考文献
[1] YIN J, SHARMA P, GORTON I, et al. Large-scale data challenges infuture power grids[C]//Proc. IEEE Int. Symp. Service-Oriented Syst. Eng., Mar. 2013, : 324-328.
[2] 中国电机工程学会信息化专委会. 中国电力大数据发展白皮书[M]. 北京: 中国电力出版社, 2013.
[3] 陈峥, 尹耕, 康晓东, 等. 基于大数据平台的电力设备运作状态监测系统研究[J]. 电力大数据, 2019(11): 1-7
[4] 梁少培. 基于大数据的电力设备运行数据分析系统的设计与实现[D]. 北京: 华北电力大学, 2018.
[5] WANG P, RAO L, LIU X, et al. D-pro: Dynamic data center operationswith demand-responsive electricity prices in smart grid[J]. IEEE Trans. Smart Grid, 2012, 3(4): 1743-1754
[6] MA H, EKANAYAKE C. Power transformer fault diagnosisunder measurement originated uncertainties[J]. IEEE Trans. Dielectr. Electr. Insul., 2012, 19(6): 1982-1990
[7] 严英杰. 基于大数据分析技术的输变电设备状态评估方法研究[D]. 上海: 上海交通大学, 2018.
[8] 尚海昆. 电力变压器局部放电信号的特征提取与模式识别方法研究[D]. 北京: 华北电力大学, 2014.
[9] 黄炳洪. 电力变压器常见缺陷及其处理[J]. 高压电器, 2003, 1(1): 78-80
[10] WERON R. Modeling and forecasting electricity loads and prices: A statistical approach[M]. New York: Wiley, 2006.
[11] PENG W, LUO X. Vibration fault diagnosis of hydro-turbinegenerating unit based on rough sets and support vector machine[J]. Trans. China Electro-Tech. Soc., 2006, 21(10): 117-122
[12] 陈佳俊. 基于大数据分析的输变电设备异常检测[D]. 上海: 上海交通大学, 2016.
[13] ZENG Y, LIANG Y. Eigenvalue-based spectrum sensing algorithmsfor cognitive radio[J]. IEEE Trans. Commun., 2009, 57(6): 1784-1793
[14] CAO Y, CAI L, QIU C, et al. A random matrix theoretical approach to early event detectionusing experimental data[J]. Journal of Latex Class Files, 2015, 13(9): 1-4
[15] 国家电网公司生产技术部. 油浸式变压器(电抗器)状态评价导则: Q/GDW 169-2008 [S]. 北京: 中国电力出版社, 2008.
[16] 中国电力工业部. 电力设备预防性试验规程: DL/T 596—2005[S]. 北京: 中国电力出版社, 2008.
[17] ZHU Y, YIN J. Study on application of multi-kernel learning relevancevector machines in fault diagnosis of power transformers[J]. Proc. CSEE, 2013, 33(22): 68-74