基于特征值分解和快速独立分量分析的谐波/间谐波检测方法

针对电力系统谐波污染问题,提出了基于特征值分解和快速独立分量分析(FastICA)的谐波/间谐波检测算法。该方法在不需要任何先验知识的情况下,将单道电力系统混合信号通过时间延迟构造出多道观测信号,对其自相关函数进行特征值分解确定原谐波/间谐波信号中源信号频率成分的个数,确定观测信号矩阵的阶数,再利用FastICA算法对谐波/间谐波信号中各个频率成分进行分离提取,借助频谱分析得到各个成分的频率估计。在此基础上,借助最小二乘法对谐波/间谐波信号进行幅值和相位估计。通过仿真实验与其他经典算法比较,充分说明了所提出算法的可行性、准确性和有效性。

A Detection Method for Harmonic/Inter-harmonic Based on Eigenvalue Decomposition and Fast Independent Component Analysis

Based on eigenvalue decomposition(EVD) and fast independent component analysis(FastICA) methods,this paper proposes a new method to detect the harmonic/inter-harmonic in power systems.First,the single-channel power system signals are transformed into multi-channel observation signals with time delay transform without any prior knowledge.By EVD of observation signal auto-relation functions,the number of independent frequency components in original harmonic/inter-harmonic is determined,which is used to correct the dimension of the observed signal matrix.Then,FastICA algorithm is adopted to separate and extract various frequency components,and frequency of each component is estimated by spectrum analysis.Finally,amplitude and phase of each component is accurately detected by least square method.Simulation results prove the feasibility,accuracy and validity of the proposed method.