基于改进S变换估计电能质量信号的噪声水平

Noise Estimation of Power Quality Signal Based on Modified S-Transform

提出了基于改进S变换估计电能质量信号噪声水平的方法。该方法根据能量聚集度的概念确定了改进S变换的时频分辨率。同时定义了与白噪声改进S变换功率谱相关的统计量,利用其分布特性检测电能质量信号能量在改进S变换时频面的分布区域,并采用迭代算法消除电能质量信号能量对噪声估计的影响,根据改进S变换时频面只含噪声区域的功率谱实现了对噪声方差的估计。仿真结果表明,当采样率达到一定程度时,采用该方法估计噪声方差有很高的可信度。

A modified S-transform （MST） based method for noise level estimation of power quality signal is proposed. According to the concept of energy aggregation degree the time-frequency resolution of MST is determined. Meanwhile, the statistic related to MST power spectrum of white noise is defined, and by use of its distribution characteristic the area of power quality signal＇s energy in MST time-frequency plane is detected. The iteration algorithm is adopted to eliminate the impact of power quality signal on noise estimation, and based on the fact that MST time-frequency plane only contains the power spectrum in noise area the estimation of noise variance is implemented. Simulation results show that the estimation result of noise variance is believable while the sampling rate achieves a certain extent.