基于原子分解快速算法的电能质量扰动检测与分类研究

发布时间：2018-04-24 10:15

本文选题：电能质量 + 扰动检测与分类　；参考：《燕山大学》2015年硕士论文

【摘要】：目前,智能化精密仪表的广泛运用,对供电质量提出了更加严格的要求;同时由于大量扰动负荷接入电网或其它扰动源存在,使得供电质量问题日益凸显。因此,研究分析电能质量扰动信号具有非常重要的意义。本文采用原子分解技术来分析各种电能质量扰动信号。为克服传统原子分解技术存在计算量过大的不足,本文将原子离散参数连续化,大大减少重构信号需要的原子数且使迭代结果更准确;针对频率范围较大的谐波、衰减振荡等扰动信号,采用快速傅里叶变换对最优原子频率进行预求解,从而降低原子库的规模;采用粒子群算法及遗传算法对匹配追踪算法进行优化。仿真算例表明,粒子群优化匹配追踪算法的性能优于遗传算法优化匹配追踪算法。采用粒子群优化的匹配追踪算法及基于扰动信号特征的连续相关原子库,对6种单一电能质量扰动(暂降信号、暂升信号、闪变信号、谐波、衰减振荡、尖峰信号)进行分析。仿真研究表明,该方法可快速准确地提取电能质量信号的扰动特征,且有较好的抗噪性能。由于连续相关原子库具有针对性,所以本文采用连续相关原子库及粒子群优化的匹配追踪算法对扰动信号进行分类。通过多重扰动信号的仿真算例验证,该分类方法可以很好的完成扰动信号的分类,并获取扰动信号的所有参数。
[Abstract]:At present, with the wide application of intelligent precision instruments, the quality of power supply is required more strictly, and the problem of power supply quality is becoming more and more serious because a large number of disturbance loads are connected to the power network or other disturbance sources. Therefore, it is very important to study and analyze the power quality disturbance signal. In this paper, atomic decomposition technique is used to analyze various power quality disturbance signals. In order to overcome the disadvantages of the traditional atomic decomposition technique, the discrete parameters of atoms are continuous, which greatly reduces the number of atoms needed to reconstruct the signal and makes the iterative results more accurate. The fast Fourier transform (FFT) is used to pre-solve the optimal atomic frequency so as to reduce the scale of atomic library. Particle swarm optimization and genetic algorithm are used to optimize the matching and tracking algorithm. Simulation results show that the performance of particle swarm optimization matching tracking algorithm is better than that of genetic algorithm. Using the matching tracking algorithm of particle swarm optimization and the continuous correlation atomic library based on the characteristic of disturbance signal, six kinds of single power quality disturbances (sag signal, hoisting signal, flicker signal, harmonic wave, attenuation oscillation, peak signal) are analyzed. The simulation results show that the proposed method can extract the disturbance characteristics of power quality signals quickly and accurately, and has good anti-noise performance. Because of the pertinence of the continuous correlation atom library, this paper uses the continuous correlation atomic library and the matching tracking algorithm of particle swarm optimization to classify the disturbance signal. The simulation results show that the classification method can achieve the classification of the disturbance signal and obtain all the parameters of the disturbance signal.
【学位授予单位】：燕山大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TM711

【参考文献】