论文部分内容阅读
电力系统本身噪声及录波装置采样引入的噪声对电力系统信号数据压缩及故障信息提取造成不利影响。传统软硬阈值降噪法未考虑噪声本身特性,一方面,其降噪处理后难以分辨故障发生的起始时刻及持续时间,不利于故障检测;另一方面,软硬阈值降噪后将保留过多冗余系数且一些有用信息被滤除,不利于数据压缩。文中给出基于尺度内和尺度间的暂态数据降噪方法,在一个确定方差的未知区间内,将小波系数建模为独立具有零均值高斯随机变量,其方差由相关区间来确定且缓慢变化,利用邻域尺度内和尺度间小波系数之间的相关性,估计噪声方差以求得无噪声系数,提取故障信息时刻并降低噪声对故障辨析带来的不利影响,得到数据量极少的高频系数,有利于故障数据压缩。
The noise of the power system itself and the noise introduced by the sampling device adversely affect the power system signal data compression and fault information extraction. The traditional hardware and software threshold denoising method does not consider the characteristics of the noise itself. On the one hand, it is difficult to distinguish the starting time and duration of the fault after the noise reduction process, which is not conducive to fault detection. On the other hand, Too much redundancy coefficient and some useful information is filtered, is not conducive to data compression. In this paper, a method of noise reduction based on the inter-scale and inter-temporal data is presented. In a unknown interval of variance, the wavelet coefficients are modeled as Gaussian random variables with zero mean respectively. The variance is determined by the relevant interval and changes slowly , The correlation between the wavelet coefficients in the neighborhood scale and the scale is used to estimate the noise variance so as to obtain the noiseless coefficient, extract the moment of the fault information and reduce the adverse effect of the noise on the fault identification, and obtain the extremely high data amount Frequency coefficient, is conducive to fault data compression.