基于最小模MUSIC算法的电力谐波/间谐波检测方法

Detection method of harmonic and inter-harmonic in power based on minimum modulus MUSIC algorithm

在电力谐波/间谐波检测中,以传统傅里叶变换为基础的分析方法分辨力差、检测精度较低,基于空间谱估计的多重信号分类(MUSIC)方法易受噪声干扰,且对各信号幅值和相位检测的准确性差等问题,文中提出对电力谐波/间谐波参数准确检测的最小模多重信号分类(MNM-MUSIC)算法,它由采样序列得到自相关函数并构造自相关矩阵,由此得到相关阵的特征值,根据最小描述长度(MDL)准则估计信号源数目,进一步用最小模方法对MUSIC算法做出修正,即利用小特征值对应的噪声子空间的线性组合与信号矢量的关系估计出各信号的频率,最后用扩展Prony方法得到各信号幅值和相位信息,由此完成对电力谐波/间谐波参数的准确检测.仿真结果表明:与经典MUSIC算法相比,此方法的优点在频率检测方面精确度更高,抗噪声能力更强,在电力谐波/间谐波分析检测中具有更高的准确性.

An analytical method based on the traditional Fourier transform is poor in resolution ratio and detection accuracy in the detection of harmonics and inter-harmonics,the method of multiple signal classification（ MUSIC） based on spatial spectrum estimation easily suffers from noise interference,and it is poor in accuracy for the detection of every amplitude and phase. A minimum modulus multiple signal classification（ MNM-MUSIC） is proposed which can accurately detect the parameter of harmonics and inter-harmonics. The advantage of this method is that the accuracy is higher in terms of frequency detection compared with the classical MUSIC algorithm. Auto correlation function can be obtained from sampling sequence and the structure self correlation matrix,so the characteristic value of the self correlation matrix can be obtained,then signal source number is estimated based on the minimum description length（ MDL） criterion,and the MUSIC algorithm is modified by the method of minimum modulus. We estimate the frequency of each signal after linear combination with the relation of the direction vector and small characteristic value correspondingly noise subspace. Finally,we get the signal amplitude and phase information by extended Prony method,so the accurate detection of power parameter is finished. Simulation results show that the method has higher precision in the detection of harmonics and inter-harmonics analysis in power systems.