基于奇异值分解的PMU数据恢复法

A PMU Data Recovery Method Based on Singular Value Decomposition

同步相量量测单元(phasor measurement units,PMUs)因其同步性、快速性和准确性成为电网动态实时监测最有效的手段之一,并得到迅速发展。但由于通信堵塞、硬件故障、传输延迟等因素,现场PMU通常存在不同程度的数据丢失问题,严重制约其在电力系统中的应用与发展,甚至威胁电网安全。该文通过分析现场PMU量测信息,归纳了3种数据丢失类型;提出基于奇异值分解的动态数据成分分析方法,并采用盖氏圆法确定其最优特征成分数,可有效提取原始信号中各特征分量。进一步,提出基于训练、验证与丢失数据分类的数据恢复迭代计算方法,实现系统动态条件下仅利用单通道PMU量测信息的高精度数据恢复。通过仿真与实测数据验证该方法的可行性和准确性。该方法解决传统方法对系统动态信号丢失数据恢复失真的问题,为PMU量测信息更好的应用于电力系统的各个方面提供保证。

Phasor measurement units(PMUs) have become the basic means of real-time monitoring and closed-loop control of power grids due to their synchronization,rapidity and accuracy,and have been rapidly developed.However,in practical applications,PMUs usually have different levels of data loss problems due to communication congestion,hardware failure,transmission delay and other factors,which seriously restrict its applications in the power systems and even threaten the security of the grid.In this paper,three types of data loss were summarized to simplify the problem by analyzing the filed PMU measurement.A dynamic data component analysis method based on singular value decomposition was proposed,the optimal feature component number is determined by Gerschgorin disk estimator,which can effectively extract the feature components in the original signal.Furthermore,an iterative calculation method based on training,verification and missing data classification was developed,which realizes high-precision data recovery using only single PMU measurement information under dynamic conditions.The feasibility and accuracy of the proposed method were verified by simulation and measured data.Compared with the traditional method,the proposed approach is able to realize a decent recovery for the data loss during dynamic process,which provides a guarantee for the PMU measurement information to be better applied to various aspects of the power system.