摘要
状态实时监测与控制对于电网安全经济运行具有重要意义。提出一种鲁棒预测辅助电力系统状态估计算法,即广义最大似然-扩展卡尔曼滤波法(GM-EKF)。该算法以预测辅助状态估计为基础,由线性回归模型构建、异常值辨识、鲁棒预白化和鲁棒滤波等过程构成。在高斯噪声分布时,GM-EKF可以有效降低异常值对估计性能的不良影响。仿真实验在多种不同测试系统中进行,且每个测试都包括正常运行、多不良数据和负荷突变3种情况,实验结果验证了GM-EKF的可行性和鲁棒性。
Real-time monitoring and control of system operation states play a significant role in the secure and economical operation of power systems.A robust forecasting-aided state estimation algorithm for power systems,called generalized maximum likelihood extended kalman filter（GMEKF）,is proposed.On the basis of forecasting-aided state estimation,GM-EKF consists of such processes as construction of a linear regression model,identification of abnormal value,robust pre-whitening noise and robust filtering.GM-EKF can effectively decrease the impact of the identification of abnormal value on estimation results if noise follows normal distribution.Simulation experiments are carried out on various systems in such cases as normal condition,multiple bad data and sudden load change,which verify the feasibility and robustness of GM-EKF.
出处
《现代电力》
北大核心
2017年第4期33-39,共7页
Modern Electric Power
关键词
预测辅助状态估计
异常值
GM估计
鲁棒
forecasting-aided state estimation
abnormal value
GM estimation
robust
