摘要
随着广域测量系统(WAMS)的广泛应用,低频振荡在线辨识成为可能,但在实际系统中存在大量的强噪声干扰,但传统的特征根分析方法对噪声较为敏感,在低频振荡辨识过程中有可能出现辨识不准确情况。鉴于互相关函数处理信号过程中不会产生新的极点,文中提出采用互相关函数对WAMS采样信号进行预处理,然后通过矩阵束算法进行辨识,进一步提高了矩阵束算法的抗噪能力与辨识准确性,为低频振荡在线辨识奠定了基础。通过理想算例和8机36节点算例仿真表明,CCF-MP算法在强噪声环境下亦有很高的辨识精度。
The online identification of low frequency oscillation became possible with a wide range applications of WAMS. However,there are much strong interference noises in the power system. In view of the traditional analysis method is very sensitive to noise,which may lead to the identification results are not accurate. Because CCF method would not generate new poles in the process of the signal preprocessing,this paper presents that processing signals base on cross-correlation function first,and then identification of vibration modal parameters base on matrix pencil. This method improved the antinoise ability and the identification accuracy of matrix pencil algorithm,which lay the foundations for the low frequency oscillation online identification. This paper sets the ideal signal and the EPRI-36 as the simulation example. The simulation results show that CCF-MP method has higher identification precision under strong noise environment.
出处
《电力系统及其自动化学报》
CSCD
北大核心
2015年第12期97-102,共6页
Proceedings of the CSU-EPSA
关键词
低频振荡
互相关函数
矩阵束
在线辨识
low frequency oscillation
cross-correlation function(CCF)
matrix pencil(MP)
online identification