电能质量扰动事件的检测和辨识算法研究

发布时间：2018-01-30 13:50

本文关键词： 电能质量扰动特征量提取 S变换支持向量机　出处：《华中科技大学》2014年硕士论文　论文类型：学位论文

【摘要】：电能是现代社会使用最为广泛的一种能源。随着科学技术的进步，电力电子，计算机，信息处理等高新技术产业的发展和普及，对电能质量也提出了更高的要求。电能质量扰动所带来的各种问题引起了供电部门和用户的关注。为了降低电能质量扰动所带来的损失，改善电能质量现状，首先需要对电能质量扰动进行辨识。本文针对电能质量扰动问题，主要在电能质量扰动和分类方面做出了相关工作，取得了如下成果：（1）本文采用改进S变换和多级支持向量机相结合的电能质量辨识方法，实现了6类单一型事件以及2类复合型电能质量事件的识别，并取得了不错的分类效果。（2）采用多个2分类的支持向量机分类器构成的电能质量识别系统，易于向系统中添加新的电能质量扰动类型，省去了复杂的人工智能方法的训练过程以及由训练样本不足而引起的分类误差。（3）基于S变换的特征提取方法抗噪声能力强，识别率高。S变换还可以采用傅立叶变换实现快速运算，，很大程度上减小特征提取的时间，在电能质量的实时监测和辨识中具有很大的应用空间。（4）提出了复合特征量的概念，将具有相似辨识功能的特征量合成为一个特征量，防止一个特征量对功能辨识的缺陷。
[Abstract]:Electric energy is the most widely used energy in modern society. With the progress of science and technology, power electronics, computer, information processing and other high-tech industries are developing and popularizing. In order to reduce the loss caused by power quality disturbance and improve the current situation of power quality, various problems caused by power quality disturbances have attracted the attention of power supply departments and users. First of all, it is necessary to identify the disturbance of power quality. In this paper, the power quality disturbance problem, mainly in the power quality disturbance and classification work has been done, and achieved the following results: In this paper, six types of single events and two types of complex power quality events are identified by using improved S transform and multilevel support vector machine. And achieved a good classification effect. The power quality recognition system based on multiple support vector machine classifiers is easy to add a new type of power quality disturbance to the system. The training process of complex artificial intelligence method and the classification error caused by insufficient training samples are eliminated. 3) the feature extraction method based on S transform has strong anti-noise ability, and the high recognition rate. S transform can also use Fourier transform to realize fast operation, which greatly reduces the time of feature extraction. It has great application space in real-time monitoring and identification of power quality. In this paper, the concept of compound feature quantity is put forward, which combines the characteristic quantity with similar identification function into a characteristic quantity to prevent the defect of function identification of a characteristic quantity.
【学位授予单位】：华中科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TM711

【相似文献】