Robust Adaptive Nonlinear Kalman Filter for Synchronous Machine Parameter Calibration

Abstract

This paper proposes a robust and adaptive nonlinear Kalman filter for synchronous machine parameter calibration. The key idea is to develop the polynomial chaos-based analysis of variance (ANOVA) method for suspicious parameter detection. ANOVA allows us to derive a set of adaptive weights that can be used to address local parameter optimality issue when performing joint state and parameter estimation. It is shown that if erroneous parameters have strong correlations, the widely used methods that augment state and parameter for joint estimation will lead to large biases. By contrast, thanks to the derived adaptive weights for the suspicious parameters, the proposed method can effectively deal with the parameter dependence, yielding much better calibration results. In addition, the robustness of the proposed method enables us to filter non-Gaussian noise. Simulations carried out on the IEEE 39-bus system validate the effectiveness and robustness of the proposed approach.