摘要
针对电力系统谐波污染及间谐波难检测问题,提出一种基于单通道独立分量分析的谐波检测算法。该算法在不需要任何先验知识的前提下,对一路电力系统混合信号进行循环平移处理构造出多路信号,然后通过主分量分析法对多路信号进行降维处理,再利用快速独立分量分析(FastICA)对谐波及间谐波各个成分进行分离,利用K均值算法根据各分量频谱进行聚类,还原到观测域对同类信号求平均值得到各谐波分量,最后分别估计各谐波分量的参数。在此基础上,针对FastICA分离出的信号有畸变问题,提出改进的FastICA算法以提高分量估计的稳定性和有效性。仿真结果表明只要各谐波分量在频谱上是合理分离的,该算法就能有效地提取出各谐波和间谐波分量。同时,改进FastICA算法具有更高的稳定性,有效性和准确性。
In allusion to the detection of harmonic pollution in power systems and the difficulty in inter-harmonic detection, a harmonic detection algorithm based on single channel independent component analysis（ICA） is proposed. With no need for any a priori knowledge, through cycle-spinning of single-channel hybrid signal of power system a multi-channel signal is formed; then by use of principal component analysis the dimensionality of obtained multi-channel signal is reduced; and then the harmonic and inter-harmonic components are separated by FastICA, and based on the frequency spectrum of each component the clustering is performed by K-means algorithm and restored to the observation domain for averaging of congeneric signal to achieve harmonic components; finally the parameters of each harmonic component are estimated. On this basis, to resolve the distortion in the signal separated by FastICA, an improved FastICA algorithm is proposed to improve the stability and validity of component estimation. Simulation results show that as long as the harmonic components are rationally separated in spectrum, harmonic and inter-harmonic components can be extracted by the proposed algorithm. Meanwhile, the improved FastICA algorithm is more stable, effective and accurate.
出处
《电网技术》
EI
CSCD
北大核心
2013年第10期2959-2964,共6页
Power System Technology
基金
四川省科技支撑计划项目(2012GZ0009)
四川省电力公司科技项目(12H1541)~~
关键词
谐波检测
间谐波检测
主分量分析
单通道独立分量分析
改进独立分量分析
harmonic detection
inter-harmonic detection
principal component analysis
single channel independent component analysis
improved ICA
