自适应无迹卡尔曼平滑算法及其在电力系统中的应用  

Adaptive Unscented Kalman Smoothing Algorithm and Its Application in Power System

作　　者：龙嘉川 王先培[1] 赵宇[1] 朱国威[1] 代荡荡 田猛[1] 

机构地区：[1]武汉大学电子信息学院,湖北省武汉市430072

出　　处：《中国电机工程学报》2015年第23期6048-6056,共9页Proceedings of the CSEE

基　　金：国家自然科学基金项目(50677047;81201154)~~

摘　　要：精确、快速的基波分量跟踪是对电网运行状态进行分析和评估的前提。提出一种基于移动窗口迭代修正策略的自适应无迹卡尔曼平滑算法(moving window based adaptive unscented Kalman smoother,MW_AUKS)。该方法兼顾稳态检测精度和动态检测速度,前者通过在前向无迹滤波过程中并行嵌入一个后向迭代修正的Rauch-Tung-Striebel平滑器实现,后者则先依据窗口内平均新息量在线判断是否有突变发生,再对状态估计协方差做自适应修正运算。利用建立的基波分量非线性状态估计模型对所提算法进行验证,结果表明所提算法可精确跟踪到基频、功率角、有功功率、视在功率等参数,并大幅提高初始收敛速度,同时准确判断和快速跟踪到状态突变。Accurate and fast tracking of fundamental components is a precondition for analysis and estimation of the power grid. A novel moving window based adaptive unscented Kalman smoother（MW_AUKS） which adopts iterative correction strategy was proposed. This method gives consideration to both the steady-state detection accuracy and dynamic detection speed. The first objective is achieved by introducing a backward Rauch-Tung-Striebel smoother to the forward unscented Kalman filter simultaneously. Based on the mean value of innovation within the window, whether sudden change of the state has happened is judged online. Then, state estimation covariance would be reset adaptively once the change is confirmed. Finally, multigroup simulations are conducted by means of a presented nonlinear state space estimation model. Results show that the new algorithm can effectively estimate the fundamental components, such as frequency, power angle and active/apparent power, and greatly speedup the rate of initial convergence. Meanwhile, accurate detection and rapid tracking of the state sudden change could be achieved.

关 键 词：无迹卡尔曼滤波 状态估计模型 自适应滤波 平滑器 基波分量 

分 类 号：TM74[电气工程—电力系统及自动化]