论文部分内容阅读
采用基于S变换下的模时频矩阵局部相似度对短时电能质量扰动进行分类。首先,由时频尺度缩放将各类持续时间不同的扰动特征标准化,按照各扰动特征选取其模时频矩阵特定部分,以此建立矩阵大小不一且特征明显的各类扰动标准化模板;然后计算未知扰动各局部模时频矩阵与各类扰动标准化模板之间的相似度,按照相似度最大原则将扰动进行分类。由于各类扰动在不同的时频区域高度聚合了自身特征,采用局部时频区域相似度原理大大提高了同类扰动的相似度,建立的各类扰动标准化模板之间差异明显,不使用辅助性分类器而直接实现扰动分类,且分类过程简单明确,可用于不同时间长度的扰动分类。该方法有效利用局部区域明确的时频相关性,抗干扰能力比采用全局相似度强。仿真结果表明了很好的短时电能质量扰动分类效果。
The short-term power quality disturbances are classified based on the local similarity of the mode-time-frequency matrix under S-transform. First of all, the perturbation features of different durations are standardized by time-scale scaling, and the specific parts of the matrix of time-frequency matrix are selected according to the perturbation features, so as to establish various perturbation standardized templates with different matrix sizes and characteristics. Then, The unknown similarity between the local time-frequency matrix and the perturbed standardized templates is unknown, and the disturbance is classified according to the principle of maximum similarity. Since all kinds of disturbances highly aggregate their own characteristics in different time-frequency regions, the similarity of similar disturbances is greatly improved by using the principle of local time-frequency region similarity, and the differences between the established types of disturbed standardized templates are obvious. The auxiliary classification And directly implement the disturbance classification, and the classification process is simple and clear, which can be used for the disturbance classification of different time lengths. The method effectively utilizes the definite time-frequency correlation in the local area, and the anti-interference ability is stronger than the global similarity. The simulation results show that the short-term power quality disturbance classification effect.