基于最大后验估计的谐波源定位

Harmonic Source Localization Based on Maximum Posteriori Estimation

在建立电网的谐波量测方程后,通过参数辨识算法估计节点谐波电流以定位谐波源,但受量测装置数和量测噪声的制约,传统参数辨识算法定位效果欠佳。本文提出一种基于最大后验估计的参数辨识算法来估计谐波电流。首先在确保量测方程能观性的前提下减少量测装置的数量,其次根据量测方程的线性关系与谐波源稀疏分布的先验知识建立谐波电流的后验概率密度函数,最后采用最大期望估计算法最大化该后验概率密度函数从而辨识谐波电流。采用Matlab软件仿真了含谐波源的IEEE33节点的系统,分析了谐波源数、量测装置数、量测噪声对算法准确度的影响,验证了本文方法的准确性和鲁棒性。

After the measurement equation for harmonics in power grid is established,the node harmonic current can be estimated by a parameter identification algorithm to locate harmonic sources.However,the traditional parameter identification algorithm is in efficient to locate due to the limited number of measurement devices and the noise of measurement.In this paper,a parameter identification algorithm based on maximum posteriori estimation is presented for harmonic current estimation.Firstly,the number of measurement devices is reduced under the premise of ensuring the observability of the measurement equation.Secondly,the posterior probability density function of harmonic current is established according to the linearity of the measurement equation and the prior knowledge of harmonic source sparse distribution.Finally,harmonic current can be identified by maximizing the posterior probability density function using the expected maximum(EM)algorithm.An IEEE33-node system with harmonic sources is simulated by Matlab,and the influences of the number of harmonic sources,the number of measurement devices and the noise of measurement on the accuracy of the algorithm are analyzed,which verifies the accuracy and robustness of the proposed method.