基于改进数学形态学与S变换的暂态电能质量扰动检测

摘要

In the traditional method,the detection of power quality disturbance signals is easily affected by thenoise and it is difficult to locate the feature value accurately. To solve these problems,an improved mathematical morphol-ogy was presented in this paper which can pre-process the noise that exists in power quality disturbance signals. The S-transform spectral standard deviation,the time squared amplitude and mean in S-matrix modulus and the contour of S-matrix modulus were selected as feature values in time and frequency domain. Six types of single disturbance and two types of complex disturbance were detected and located. Compared with traditional mathematical morphology and Kalman filtering,it is found that the improved mathematical morphology can eliminate the noise effectivelyand strengthen the useful signal. The disturbance characteristics in time and frequency domain can be reflected by the se-lected features of S-transform,which makes the detection and localization results of power quality disturbance signal more accurate.%针对电能质量扰动信号检测受噪声影响大且特征值定位不准确的问题，提出一种改进数学形态学方法对电能质量扰动信号中存在的噪声干扰进行预处理，并选取 S 变换频谱标准差、S 模矩阵时间幅值平方和均值以及 S模矩阵等高线这3种时频域的特征值，实现了6种单一扰动信号和2种复合扰动信号的检测和定位。仿真结果表明，与传统的数学形态学滤波和卡尔曼滤波方法相比，改进的数学形态学方法去噪效果好，而且可以加强原信号中的有用信号，S变换中选取的特征值从时域和频域共同体现扰动特征，使电能质量扰动信号的检测定位更加准确。