混合电能质量扰动的检测与分类

发布时间：2018-03-18 22:42

本文选题：混合电能质量扰动　切入点：CEEMD　出处：《西南交通大学》2015年硕士论文　论文类型：学位论文

【摘要】：随着国民经济的飞速发展,电力电子设备得到广泛的应用,各种冲击性、非线性、波动性的负载也大幅增加,导致了一系列的电能质量问题,使得电能质量污染日趋严重。同时,电力用户对电力的可靠性及质量的要求都在不断提高,许多精密仪器对电能质量越来越敏感,电能质量的下降,将对人们的生活生产造成很大的危害。因此,准确的对电能质量扰动参数进行检测、识别电能质量扰动类别,对于电能质量污染的抑制和治理具有重要意义。本文在前人研究的基础上,对电能质量扰动检测和混合电能质量扰动的分类问题进行了分析。在电能质量扰动参数检测方面,本文分别采用了CEEMD、IELMD和HVD这三种方法对电能质量扰动进行检测。方法一,采用CEEMD对电能质量扰动进行分解,对所得到的IMF分量进行Hilbert变换得到原始信号的瞬时频率和瞬时幅值。仿真结果表明,该方法对5种单一扰动和3种简单的混合扰动都具有较好的检测效果。方法二,采用IELMD方法将电能质量扰动信号分解成多个PF分量,对得到的PF分量进行Hilbert变换得到对应的瞬时频率。仿真结果表明,IELMD方法对5种常见单一扰动具有较好的检测效果。方法三,通过对扰动信号进行HVD分解,得到扰动的瞬时频率和瞬时幅值。仿真结果表明,HVD分解对谐波和间谐波具有较好的检测效果。在暂态电能质量扰动分类方面,根据谱峭度的定义和Cohen类时频分布的特点,将基于CWD的谱峭度与有效值这两种算法结合起来应用于暂态电能质量扰动分类,仿真结果表明,该算法具有良好的抗噪性能,不仅对单一扰动的分类精度较高,对简单的双重扰动的分类效果也具有较高的分类精度。除此之外,该算法直接通过阈值分类,不需要采集大量的训练数据对分类器进行训练,具有较好的可行性。在混合电能质量扰动分类方面,根据混合电能质量扰动在时域和频域的特点,将多种时频分析算法融合起来提取多个特征,采用决策树对混合电能质量扰动进行分类。该算法从基频幅值、频率点个数、瞬时分量等多个角度出发,采用CEEMD、不完全S变换、动态测度联合提取了9个特征量,考虑到混合扰动中单一扰动之间的相互干扰,对9个特征量的进行合理的组合,设计了一个决策树对混合扰动进行识别分类。通过仿真数据和实测数据对算法进行验证,表明该算法对于单一扰动和混合扰动都具有较高的识别精度。
[Abstract]:With the rapid development of the national economy, power electronic equipment has been widely used. The load of impact, nonlinearity and volatility has increased significantly, resulting in a series of power quality problems. At the same time, the reliability and quality requirements of electric power users are constantly improving. Many precision instruments are more and more sensitive to power quality, and the power quality is declining. Therefore, accurate detection of power quality disturbance parameters, identification of power quality disturbance types, It is of great significance to restrain and control the power quality pollution. On the basis of previous studies, this paper analyzes the problems of power quality disturbance detection and hybrid power quality disturbance classification. In the aspect of power quality disturbance parameter detection, In this paper, three methods, CEEMD IELMD and HVD, are used to detect the disturbance of power quality. Method 1, CEEMD is used to decompose the disturbance of power quality. The instantaneous frequency and amplitude of the original signal are obtained by the Hilbert transform of the obtained IMF component. The simulation results show that the proposed method is effective for the detection of five single disturbances and three simple mixed disturbances. The IELMD method is used to decompose the power quality disturbance signal into several PF components, and the corresponding instantaneous frequency is obtained by Hilbert transformation of the PF component. The simulation results show that the proposed method has a better detection effect on five common single disturbances. The instantaneous frequency and amplitude of the disturbance are obtained by HVD decomposition of the disturbance signal. The simulation results show that the decomposition has a good effect on the detection of harmonics and interharmonics. According to the definition of spectral kurtosis and the characteristics of time-frequency distribution of Cohen class, two algorithms based on spectral kurtosis and effective value based on CWD are combined to classify transient power quality disturbances. The simulation results show that the algorithm has good anti-noise performance. Not only the classification accuracy of single disturbance is high, but also the classification effect of simple double disturbance is higher. In addition, the algorithm directly classifies by threshold value, and does not need to collect a large amount of training data to train the classifier. In the classification of hybrid power quality disturbances, according to the characteristics of hybrid power quality disturbances in time domain and frequency domain, a variety of time-frequency analysis algorithms are fused to extract multiple features. The decision tree is used to classify the hybrid power quality disturbance. The algorithm extracts nine characteristic quantities from the angle of fundamental frequency amplitude, frequency point number, instantaneous component and so on, using CEEMD, incomplete S transform and dynamic measure. Considering the mutual interference between the single disturbance in the mixed disturbance, a decision tree is designed to identify and classify the mixed disturbance. The algorithm is verified by the simulation data and the measured data. It is shown that the algorithm has high recognition accuracy for both single and mixed perturbations.
【学位授予单位】：西南交通大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TM711

【参考文献】