摘要
针对基于有限长单位冲激响应(finite impulse response,FIR)数字滤波器的同步相量测量原理,首先从理论上推导基波、谐波和噪声引起的测量误差与滤波器频响特性的关系,由此定义滤波器多项技术指标,并通过建立总向量误差和频率测量误差与这些技术指标的函数关系,制定实现同步相量和频率测量的FIR数字滤波器组的最优设计准则。其次,针对离散傅里叶变换(discrete Fourier transform,DTF)类动态相量测量原理,推导这类算法的等效数字滤波器组,根据最优设计准则求解优化问题,进行滤波器自由参数最优选取,实现滤波器的最优设计,并给出长度分别为2、3和4周波的相量测量滤波器,以及4和5周波的频率测量滤波器组的设计结果。最后,将同长度最优设计算法与相关性能评估中表现优越的4周波TWLS算法进行性能比较,仿真实验表明,两者在静态和动态各类测试中,总体均表现出非常接近的优越性能,而在抗谐波性能方面,优化算法具有明显优势。
Based on the study of synchrophasor measurement unit（PMU） measuring principles of finite impulse response（FIR） filter-bank, we theoretically derived the quantitative relations between synchrophasor and frequency measurement errors, i.e. TVE and FE due to fundamental off-nominal frequency, harmonics and wideband noise, with frequency responses of filter bank. By defining a group of specifications characterizing the filter bank, we built TVE and FE limits as functions of these specifications, and therefore propose optimization criteria for optimal filter design to achieve best overall measurement performance. In order to realize the optimization, we generalized DFT-based dynamic phasor measurement algorithm and derive its equivalent filter bank with free parameters. The DFT-based filters are designed by solving the optimization problems for optimal free parameters and the optimal results were given for synchrophasor measurement filters with 2-cycle, 3-cycle and 4 cycle length, and frequency measurement filters with 4-cycle and 5-cycle length. Simulation results were presented for performance comparison between well-known 4-cycle TWLS algorithm and our optimization algorithm of the same length, which shows that the two algorithms have almost the same excellent performances in all aspects, except in harmonic rejection where our optimized algorithm achieve significantly superior performance.
出处
《中国电机工程学报》
EI
CSCD
北大核心
2017年第22期6691-6699,共9页
Proceedings of the CSEE
关键词
最优滤波器设计
相量测量单元
频率测量
离散傅里叶变换测量算法
optimal filter design
synchrophasor measurement unit
frequency measurement
DFT based measurement algorithm
