摘要
变压器是电力系统的枢纽设备之一,基于油色谱的在线监测技术可以实时获取油中溶解气体含量的时间序列,从而实现对变压器的故障预警和辨识。在利用时间序列进行分析时,选择过多或者过少的数据均会导致分析结果不准确。因此,该文提出了基于相空间重构的油色谱分析数据最优长度选择方法。首先,利用李雅普诺夫指数验证时间序列是否具有混沌特性,对于具有混沌特性的时间序列数据,利用延迟坐标法重构其相空间,然后利用C-C法计算嵌入维数和延迟时间,在获得嵌入维数的基础上,通过分析嵌入时间序列的关联积分的收敛性,得到时间序列长度与嵌入维数之间的关系,从而可以得到数据的最优长度。最后,该文给出了计算变压器油色谱数据最优长度的流程,并利用获取到的现场的案例进行验证,验证结果表明:使用最优油色谱数据计算得到的结果与原来的计算结果一致,但是计算量和计算时间显著减少,计算效率显著提升。
Power transformer is one of the key electrical apparatus in power system, the fault of transformer can be warned and identified based on the time series of dissolved gas in oil, which is acquired by the DGA on-line monitoring technology in real time. More or less data will cause the result inaccurate, when using DGA time series to analysis. Therefore, this paper proposed an optimal length selection method of DGA data based on phase space reconstruction. Firstly, the Lyapunov exponent is used to verify whether the time series is chaotic, and using the delay coordinate method to reconstruct the phase space for the time-series data with chaotic characteristics. Then, the embedding dimension and the delay time can be calculated by using C-C method. After that, the relationship between the time series length and the embedding dimension can be obtained by analyzing the convergence of the correlation integral of the embedded time series, so that the optimal length of the data can be obtained. Finally, the flow of calculating optimal length of DGA data is given, and the DGA data in field is used to verify the method. The results show that the warning values obtained by using the optimal DGA data are consistent with the result by using all the data, but the computation and computing time are significantly reduced and the computational efficiency is improved significantly.
作者
齐波
张鹏
荣智海
李成榕
杨祎
陈玉峰
QI Bo;ZHANG Peng;RONG Zhihai;LI Chengrong;YANG Yi;CHEN Yufeng(State Key Laboratory of Alternate Electrical Power System with Renewable Energy Sources (North China Electric Power University), Changping District, Beijing 102206, China;Shandong Electric Power Research Institute of SGCC, Jinan 250003, Shandong Province, China)
出处
《中国电机工程学报》
EI
CSCD
北大核心
2018年第8期2504-2512,共9页
Proceedings of the CSEE
基金
国家高技术研究发展计划项目(863计划)(2015AA050204)~~
关键词
油中溶解气体分析
数据最优长度
相空间重构
C-C方法
dissolved gas analysis
optimal length of time series
phase space reconstruction
C-C method