基于无迹卡尔曼滤波的电力系统动态状态估计研究

发布时间：2018-06-07 10:23

本文选题：状态估计 + 无迹卡尔曼滤波　；参考：《华北电力大学》2014年硕士论文

【摘要】：电力系统状态估计是电力系统状态监测与控制的核心。状态估计的目的在于根据测量和网络模型获得电力系统准确的实时运行状态，以便进行电力系统分析、预测或控制等，提高系统的安全与经济运行水平。目前实际电力系统中常用的方法是基于最小二乘法的静态状态估计，静态状态估计的缺陷是根据某一时间断面获得的系统状态，表征的是系统的稳态运行状况。基于卡尔曼滤波原理的动态状态估计方法是为了进一步提高实时性而提出的，，具有重要的研究价值。本文介绍了卡尔曼滤波的基本原理及几种基于卡尔曼滤波原理的滤波方法，其中无迹卡尔曼滤波作为一种非线性滤波方法，并没有对非线性系统进行线性化，而是通过无迹变换这一方法对非线性系统的均值和方差信息进行传递，该方法与扩展卡尔曼滤波方法相比具有更好的数值稳定性。本文基于无迹卡尔曼滤波方法，结合相量测量单元在电力系统中的应用，对电力系统动态状态估计计算方法进行研究。系统模型中预测模型不可能完全精确，参数估计会有误差，系统噪声的统计特性亦不可知，因此引入渐消记忆指数加权的噪声统计估值器，既可以估计时变系统噪声，亦可将模型误差归入噪声中进行估计。引入噪声统计估值器后，即得到了自适应无迹卡尔曼滤波方法。对于扩展卡尔曼滤波、无迹卡尔曼滤波、Cubature卡尔曼滤波和自适应无迹卡尔曼滤波方法，均在MATLAB中编程实现，并结合IEEE14、IEEE30、IEEE57和IEEE118测试系统，对各种方法动态状态估计性能进行比较，自适应无迹卡尔曼滤波取得了较好的估计效果，与原有的算法进行比较，验证了算法的有效性。
[Abstract]:Power system state estimation is the core of power system state monitoring and control. The purpose of state estimation is to obtain accurate real-time operation state of power system according to measurement and network model, so as to analyze, predict or control power system, and improve the security and economic operation level of power system. At present, the commonly used method in power system is static state estimation based on least square method. The defect of static state estimation is the system state obtained according to a certain time section, which represents the steady state of the system. The dynamic state estimation method based on Kalman filtering principle is proposed to further improve the real-time performance, which has important research value. This paper introduces the basic principle of Kalman filter and several filtering methods based on Kalman filter principle. As a nonlinear filtering method, unscented Kalman filter does not linearize the nonlinear system. The unscented transformation is used to transmit the mean and variance information of nonlinear systems. The method has better numerical stability than the extended Kalman filtering method. Based on the unscented Kalman filtering method and the application of phasor measurement unit in power system, the calculation method of power system dynamic state estimation is studied in this paper. The prediction model in the system model cannot be completely accurate, the parameter estimation will have errors, and the statistical characteristics of the system noise will not be known. Therefore, a noise statistical estimator weighted by fading memory exponents can be used to estimate the time-varying system noise. The model error can also be classified into noise for estimation. The adaptive unscented Kalman filter is obtained by introducing the noise statistical estimator. For extended Kalman filter, unscented Kalman filter cuboid Kalman filter and adaptive unscented Kalman filter, they are all programmed in MATLAB. The dynamic state estimation performance of various methods is compared with IEEE 14, IEEE30, IEEE57 and IEEE118 test system. The adaptive unscented Kalman filter has achieved a good estimation effect, and compared with the original algorithm, the validity of the algorithm is verified.
【学位授予单位】：华北电力大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TM732;TN713

【引证文献】