Abstract: Voltage sag has become one of the most serious problems of power quality. For voltage sag compensation, it is critical to estimate voltage sag parameters fast and exactly. A wavelet based method for sag detection is presented in this paper. Zero-crossing point of voltage wave is detected, and 1/4 cycle of voltage signal is sampled from the zero-crossing point. The signal samples are preprocessed, and its detailed coefficients are obtained by discrete wavelet transform. It reconstructs the detailed coefficient vector and then gets the high-frequency coefficient vector （d1） with the same length of the sampling signal. The voltage sag parameters can be determined; the appearing instant of modulus maximum value of d1 is the starting point of the sag, and the modulus maximum value reflects the sag amplitude. This method is simple and easy to implement, moreover, the sag delay is less than 2.5 ms, which meets the real-time requirement of dynamic voltage restorer.

Key words: power quality, voltage sag, wavelet transform, high-frequency coefficient vector, modulus maximum

摘要： 电压暂降是最严重的电能质量问题之一，准确、快速地检测电压暂降特征值是实现电压暂降有效补偿的前提。提出了一种基于离散小波变换的电压暂降实时检测方法。该方法从电压过零点开始对电压信号进行1/4个工频周期的采样，对采样信号作预处理，进行离散小波变换获取高频系数向量，并重构高频系数向量得到与采样信号等长的高频系数向量d1，通过d1的模极大值及其极值出现对应的时刻，确定电压暂降幅值及起始时刻。该方法原理简单，易于实现，暂降时延不大于2.5 ms，满足动态电压恢复器对暂降检测实时性的要求。

关键词: 电能质量, 电压暂降, 小波变换, 高频系数向量, 模极大值