基于小波和改进神经树的电能质量扰动分类

发布时间：2018-03-22 10:02

本文选题：电能质量　切入点：扰动分类　出处：《电力系统保护与控制》2014年24期 　论文类型：期刊论文

【摘要】：准确地识别和分类电能质量扰动对分析和综合治理电能质量问题具有重要意义。提出了一种基于小波和改进神经树的电能质量扰动分类方法。该方法利用小波分解扰动信号到各个频带,在基频频带、谐波频带和高频带上分别计算其能量值和小波系数熵作为特征值,另计算基波频带扰动过程的均方根作为特征的补充,融合能量值、熵和均方根值作为扰动判断的特征向量,规范化后输入到改进神经树分类器进行训练和分类。改进神经树分类器是由神经网络和决策树及其分类规则构成。仿真表明,该方法提取特征值的计算量小且融合后的特征向量能够很好地体现不同扰动信号之间的差异信息,构造的改进神经树分类器结合了神经网络和决策树在模式分类中各自的优点,结构简单且表现出良好的收敛性、全局最优性和泛化性,分类准确率较高,能够有效地识别七种常见的电能质量扰动。
[Abstract]:Accurate identification and classification of power quality disturbances is of great significance for analyzing and synthetically treating power quality problems. A new power quality disturbance classification method based on wavelet and improved neural tree is proposed in this paper. The method uses wavelet division to classify power quality disturbances. Unperturbed signal to each frequency band, In the base frequency band, the harmonic band and the high frequency band respectively calculate the energy value and the wavelet coefficient entropy as the eigenvalues. In addition, the root mean square of the fundamental frequency band perturbation process is calculated as the supplement of the characteristic, and the energy value is fused. Entropy and root-mean-square value are used as characteristic vectors of disturbance judgment, and normalized input to the improved neural tree classifier for training and classification. The improved neural tree classifier is composed of neural network, decision tree and its classification rules. The simulation results show that the improved neural tree classifier is composed of a neural network, a decision tree and its classification rules. In this method, the computation of the extracted eigenvalues is small and the fused feature vectors can well reflect the difference information between different disturbance signals. The improved neural tree classifier combines the advantages of neural network and decision tree in pattern classification, and the improved neural tree classifier combines the advantages of neural network and decision tree in pattern classification. The structure is simple and shows good convergence, global optimality and generalization, high classification accuracy, and can effectively identify seven common power quality disturbances.
【作者单位】：湖南大学电气与信息工程学院;江西省电力科学研究院;湖南湖大华龙电气与信息技术有限公司;
【基金】：江西省电力公司科技项目(赣电科201350617)
【分类号】：TM711

【参考文献】