基于自适应插值强跟踪扩展卡尔曼滤波的电力系统动态状态估计研究

Dynamic State Estimation of Power System Based on Adaptive Interpolation Strong Tracking Extended Kalman Filter

针对扩展卡尔曼滤波(extended Kalman filter,EKF)算法在电力系统状态估计时存在鲁棒性差,精度被非线性系统的非线性程度制约大等缺点,提出一种自适应插值强跟踪扩展卡尔曼滤波(adaptive interpolation strong tracking extended Kalman filter,AISTEKF)算法,用于电力系统的动态状态估计。新算法利用自适应插值在两个连续采样点之间增加伪量测值,减小了EKF的线性化误差,有效提高了算法估计的精度;此外,该方法在EKF算法基础上引入强跟踪理论,增强了算法估计的鲁棒性。为验证所提出方法的有效性,分别运用EKF算法、自适应插值扩展卡尔曼滤波(adaptive interpolation extended Kalman filter,AIEKF)算法和AISTEKF算法对IEEE-5节点系统和IEEE-30节点系统进行动态状态估计。实验结果表明,与EKF和AIEKF算法相比,无论在高斯噪声环境下还是3种有偏噪声环境下,AISTEKF算法的电压幅值估计精度和电压相角估计精度都有显著性提高。所提出的新算法是一种鲁棒性好且估计精度高的电力系统状态估计方法。

In power system state estimation,the extended Kalman filter(EKF)algorithm is poor in robustness and is greatly restricted by the nonlinear degree of the nonlinear system in accuracy.For this reason,this paper proposes an adaptive interpolation strong tracking EKF(AISTEKF)algorithm for the dynamic state estimation of the power system.The new algorithm increases some pseudomeasurements between two continuous sampling points by using the adaptive interpolation,reducing the linearization errors of the EKF and improving the estimation accuracy of the algorithm effectively.In addition,the strong tracking theory is introduced based on the EKF algorithm to enhance the robustness of the algorithm estimation.To verify the effectiveness of the proposed algorithm,the dynamic state estimations in the IEEE-5 node system and the IEEE-30 node system are carried out by using the EKF algorithm,the adaptive interpolation EKF(AIEKF)algorithm and the AISTEKF algorithm respectively.Experimental results show that,compared the EKF with the AIEKF algorithms,the estimation accuracies of the voltage amplitude and the voltage phase Angle of the AISTEKF algorithm are improved significantly in both the Gaussian noise environment and the other three kinds of biased noise environments.The proposed algorithm is an excellent method for the power system state estimation with good robustness and high estimation accuracy.