当前位置: X-MOL 学术Int. J. Electr. Power Energy Sys. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Extended kernel Risk-Sensitive loss unscented Kalman filter based robust dynamic state estimation
International Journal of Electrical Power & Energy Systems ( IF 5.2 ) Pub Date : 2022-12-28 , DOI: 10.1016/j.ijepes.2022.108898
Wentao Ma, Xiao Kou, Junbo Zhao, Badong Chen

The traditional unscented Kalman filter (UKF) with mean square error (MSE) criterion for dynamic state estimation (DSE) is sensitive for unknown non-Gaussian noise and outliers. Leading to biased state estimates. This paper proposes a novel robust UKF with extended kernel risk-sensitive loss (EKRSL) for DSE considering unknown non-Gaussian process and measurement noises. Instead of MSE criterion, a novel robust EKRSL via the generalized Gaussian density is defined in KRSL framework, and we further develop a new robust UKF using the EnKRSL(called EKRSL-UKF). To obtain the recursive form of EKRSL-UKF, the statistical linear regression model is used and the fixed-point iteration is further utilized to iteratively get the optimal state estimate. An error constrained method is also introduced to restrict the error to address the numerical instability problem caused by large outliers. Furthermore, an enhanced EKRSL-UKF is established by using an exponential function of innovation to improve the estimation accuracy in the presence of noise uncertainties. Numerical results carried out on the IEEE 39-bus test system demonstrate that the proposed method can achieve desired robustness without loss of estimation accuracy under various conditions.



中文翻译:

基于扩展内核风险敏感损失无味卡尔曼滤波器的鲁棒动态状态估计

具有用于动态状态估计 (DSE) 的均方误差 (MSE) 准则的传统无味卡尔曼滤波器 (UKF) 对未知的非高斯噪声和异常值很敏感。导致有偏差的状态估计。本文针对未知的非高斯过程和测量噪声,为 DSE 提出了一种具有扩展内核风险敏感损失 (EKRSL) 的新型鲁棒 UKF。在 KRSL 框架中定义了一种通过广义高斯密度的新型鲁棒 EKRSL,而不是 MSE 准则,并且我们使用 EnKRSL(称为 EKRSL-UKF)进一步开发了一种新的鲁棒 UKF。为了得到EKRSL-UKF的递推形式,利用统计线性回归模型,进一步利用不动点迭代迭代得到最优状态估计。还引入了误差约束方法来限制误差,以解决由大异常值引起的数值不稳定问题。此外,通过使用创新的指数函数建立增强的 EKRSL-UKF,以提高存在噪声不确定性时的估计精度。在 IEEE 39 总线测试系统上进行的数值结果表明,所提出的方法可以在各种条件下实现所需的鲁棒性而不会损失估计精度。

更新日期:2022-12-28
down
wechat
bug