基于小波分析的电力系统低频振荡研究

发布时间：2018-04-16 08:04

本文选题：小波分析 + Prony算法　；参考：《哈尔滨工业大学》2015年硕士论文

【摘要】：我国跨省、跨区域电网的迅速发展促进了电力能源的合理分配,但由于区域机组之间的网络形式,长距离交直流电力输送,以及发电机快速励磁的存在,致使系统运行中负阻尼问题凸显,导致电网中多次出现低频振荡(Low Frequency Oscillation),直接对各电厂的稳定运行造成了威胁,给国民经济的发展带来了严重破坏,因此研究电力系统的低频振荡具有重要的意义。本文基于小波理论与Prony算法对电力系统低频振荡时的电压和功率信号进行分析,为电力系统隐极同步发电机低频振荡主导模式和阻尼分析提供可靠的理论及工程实际依据。首先,分析了低频振荡的产生条件及信号特征。由于电力系统模型复杂,很难建立起整个电力系统的动态运行数学模型,本文从局部发电机运行状态和输出信号入手,分析了实际电力系统信号的非平稳性和随机性特点。考虑到小波分析理论在信号时频领域分析、非平稳信号的去噪滤波和奇异值点敏感的检测等方面具有良好的效果,因此先对数字滤波器和小波基本理论进行研究。其次,分析了电力工程系统中的单机无穷大系统模型和隐极同步发电机的低频振荡机理,结合发电机的电、磁、机械统一特性,分析出发电机的转子和定子之间的力矩不平衡导致的转子频率摆动和励磁调节器负阻尼是低频振荡的产生的根本原因。然后,以在线辨识的Prony算法为理论依据分析出电力系统低频振荡信号的特点。通过Matlab仿真验证了Prony算法对非平稳信号估计时,容易产生拟合阶次高和易受噪声影响的缺点,进而结合小波去噪滤波理论和奇异值点分析原理对电力系统低频振荡信号进行分析,并利用Matlab仿真验证算法的可行性。同时经过实验方法找到基于coif5小波基函数对发电机输出信号滤波去噪分析的有效性。在工程上利用新疆哈密电厂的实际低频振荡数据,验证了两种算法结合的正确性。最后,介绍了发电机稳定运行在线监测装置的软硬件实现。具体装置实现过程:在GPS统一授时装置信号的触发下,系统采集发电机母线电压和电流信号并通过AD转化,经过DSP与CPLD的简单运算处理,数据通过光纤通信传送给上位机,利用后台上位机VC++编程实现小波分析和Prony算法的结合得出信号的振幅、频率、相位和衰减因子,为发电机运行可视化界面显示提供准确的显示信息,指导工作人员实时调控同步发电机的运行状态。实践证明在线装置实用性强,能够解决分析电力系统低频振荡的难题。
[Abstract]:The rapid development of cross-provincial and cross-regional power grids in China has promoted the rational distribution of power sources. However, because of the network form between regional units, the long distance AC / DC power transmission and the rapid excitation of generators,As a result, the problem of negative damping in the operation of the power system is prominent, and the low Frequency oscillation occurs many times in the power network, which directly threatens the stable operation of the power plants and brings serious damage to the development of the national economy.Therefore, it is of great significance to study the low frequency oscillation of power system.In this paper, based on wavelet theory and Prony algorithm, the voltage and power signals of low-frequency oscillation in power system are analyzed, which provides a reliable theoretical and engineering practical basis for the dominant mode and damping analysis of low-frequency oscillation of power system hidden pole synchronous generator.Firstly, the generating conditions and signal characteristics of low frequency oscillation are analyzed.Because the power system model is complex, it is difficult to establish the dynamic operation mathematical model of the whole power system. In this paper, the non-stationary and stochastic characteristics of the actual power system signal are analyzed from the operation state and output signal of the local generator.Considering that wavelet analysis theory has a good effect in signal time-frequency analysis, non-stationary signal de-noising filtering and singular-value point sensitive detection, the digital filter and wavelet theory are studied first.Secondly, the single machine infinite system model and the low frequency oscillation mechanism of the hidden pole synchronous generator are analyzed, and the unified characteristics of electric, magnetic and mechanical generator are combined.It is analyzed that the rotor frequency swing and the negative damping of the excitation regulator caused by the torque imbalance between the rotor and the stator of the generator are the fundamental causes of the low frequency oscillation.Then, the characteristics of low frequency oscillation signal of power system are analyzed based on the Prony algorithm of online identification.The Matlab simulation proves that the Prony algorithm is easy to produce the shortcomings of high fitting order and easy to be affected by noise when it is used to estimate non-stationary signals.Combined with wavelet denoising filtering theory and singular value point analysis theory, the low frequency oscillation signal of power system is analyzed, and the feasibility of the algorithm is verified by Matlab simulation.At the same time, the effectiveness of filtering and denoising analysis of generator output signal based on coif5 wavelet basis function is found by experimental method.The actual low frequency oscillation data of Hami Power Plant in Xinjiang are used to verify the correctness of the two algorithms.Finally, the hardware and software implementation of the on-line monitoring device for generator stable operation is introduced.Under the trigger of GPS unified timing device signal, the system collects generator bus voltage and current signals and transforms them through AD. After the simple operation of DSP and CPLD, the data is transmitted to the upper computer through optical fiber communication.Combining wavelet analysis with Prony algorithm, the amplitude, frequency, phase and attenuation factor of the signal can be obtained by using VC programming of background PC, which can provide accurate display information for the visual interface display of generator operation.Guide the staff to adjust and control the running state of synchronous generator in real time.Practice shows that the on-line device is practical and can solve the problem of analyzing low frequency oscillation in power system.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TM712

【参考文献】