基于最优线性滤波的PMU测量Ⅱ:P类算法设计

PMU Measurement Based on Optimal Filtering Ⅱ:P Class Filter Design

系列第一篇论文详细阐述PMU最优算法设计的基础理论和最优滤波器的设计准则。该文在此基础上，研究P类测量最优算法的具体设计。具体地，针对P类测量在快速响应前提下的高准确度需求，选用天然具有谐波抑制性能的基于离散傅里叶变换（discrete Fourier transform，DFT）的动态相量测量算法，通过对其形式的扩展，建立原型滤波器，提取出矩形卷积窗长（Mw）、DFT个数（印和泰勒模型阶数（K）作为优化的自由参数。以数据上报率50fps为例，完成固定系数和变系数测量滤波器组的最优化设计，并构成完整的P类测量优化算法。通过仿真试验，评估优化算法的实际测量性能，并与加窗DFT、动态同步相量测量算法（dynamicphasorestimatealgorithm，DPEA）和泰勒模型加权最小二乘算法（Taylor-modelweightedleastsquares，TWLS）等常用算法进行比较，验证所提出优化算法的性能优势。

The optimal filter design criteria for PMU measurement and its theoretical basis are derived and thoroughly discussed in Part I of the three-part paper set. Based on the optimal design criteria, Part II deals with the practical design of the optimal algorithms for P class measurement. To meet the P class requirements of high precision and very short response time, the form of the discrete Fourier transform （DFT） based dynamic algorithm was extended and select as the prototype filter, which is intrinsically immune to harmonics. The free parameter set to be optimized consists of Nw （the length of rectangular self-convolutional window）, d （the number of DFT values） and K （the Taylor series order）. The optimization results of both constant efficient filter bank and frequency-tracking based time variant filter banks （with sub-band number Q=5 and 9） are provided, where each filter bank constructs a complete PMU measurement algorithm. The excellent performance of the optimal algorithms are verified with simulation results, along with its comparison with traditional algorithms, including windowed DFT, dynamic phasor estimate algorithm （DPEA） and Taylor-model weighted least squares （TWLS） algorithms.