基于实测数据的电网频率信号特征分析与提取方法

Feature Analysis and Extraction Method of Power Grid Frequency Signal Based on Measured Data

随着新能源大规模接入,电力系统不同区域的转动惯量、频率特性、支撑能力发生了较大变化,提取与辨识频率特征可为电网特性认知以及进一步的惯量评估、频率控制、网络安全等分析提供基础。文中基于源-网-荷全景同步测量系统的大量实测数据,分析了频率与交流电网结构相关的分群现象,提出了基于皮尔逊相关系数的频率空间相关性辨识方法;提出了基于卷积神经网络的“频率指纹”提取方法,将电网特性在频域上进行了高纬特征提取;进一步,对北京、长治等10个城市的实测频率信号进行了测试与分析,给出了识别精度,验证了所提方法的有效性,为后续电力系统频率特性分析、惯量评估、网络攻击识别提供了基础。

With the large-scale integration of renewable energy,the rotational inertia,the frequency characteristics,and the support capacity of different regions in the power system have changed greatly.The extraction and identification of frequency features can provide a basis for the recognition of power grid characteristics and further analysis of inertia evaluation,frequency control,and network security.Based on a large number of measured data from the source-grid-load full-view synchronous measurement system,this paper analyzes the clustering phenomenon related to the frequency and theAC power grid structure,and proposes a frequency spatial correlation identification method based on the Pearson correlation coefficient.A“frequency fingerprint”extraction method based on convolutional neural networks is proposed to extract the high-latitude features of the power grid characteristics in the frequency domain.Furthermore,the measured frequency signals of ten cities,such as Beijing and Changzhi,are tested and analyzed,and the identification accuracy is given,which verifies the effectiveness of the proposed method and provides a basis for the subsequent frequency characteristic analysis,inertia evaluation and cyber attack identification of the power system.