基于递推预测误差最小的时变电力谐波估计

Time-Varying Power Harmonics Estimation Based on Recursive Prediction Error Minimization

提出了递推预测方法,可同时快速地跟踪各谐波分量的频率、幅值和相位。基于瞬时频率的概念,推导出可用于描述时变谐波信号的动态系统方程,其中幅值和相位是状态变量,而频率是时变参数变量。利用Newton类优化过程并通过递推最小化预测误差,所有状态和参数变量可以联合估计。大量数值仿真算例证实了该方法的有效性。

This paper proposed a method which can track the frequency, phase and amplitude of each sinusoidal component in the power signal, simultaneously and instantly. Based on the definition of instantaneous frequency, a dynamic system is deduced to describe a time-varying harmonic signal, where amplitudes and phases are its state variables and frequencies are its time-varying parameters. And then, both the states and parameters were estimated jointly by recursively minimizing the prediction error with a Newton-type optimization procedure. Numerical examples supporting the performance of the proposed algorithm were presented.