S-transform has fixed time-frequency resolution, leading to poor results of localizing transient power quality disturbances. A new method to localize the disturbances is proposed based on generalized S-transform. The method detects mutation peak in the high frequency time-amplitude curve, in order to increase localization accuracy. At first, modulus time-frequency matrix is calculated by generalized S-transform, then the disturbances' start-stop time is localized using the high frequency time-amplitude curve, and four identification features are extracted according to maximum frequency spectrum curve, fundamental frequency amplitude curve and the localization results. At last, automatic classification of disturbance signals is performed by use of a rule-based decision tree. Simulation results show that the proposed localization method is simple and intuitive, with high accuracy. The number of identification features is small and they are effective with good classification results.This work is supported by National Natural Science Foundation of China (No. 51077137).%S变换由于时频分辨率固定,从而导致定位暂态电能质量扰动的效果差.提出一种基于广义S变换的扰动定位新方法,利用高频处时间幅值曲线的突变点峰值进行定位检测,以提高扰动的定位精度.首先通过广义S变换得到扰动信号的模时频矩阵,然后利用高频处时间幅值曲线定位扰动的起止时刻,再根据最大频谱曲线、基频幅值曲线与定位结果提取四个识别特征量,最后基于分类规则树方法实现扰动信号的自动分类.仿真结果表明,所提出的定位方法简单直观,精度较高；提取的识别特征量少而有效,分类效果良好.
展开▼
