基于马氏距离的电力系统两阶段鲁棒状态估计方法研究

Mahalanobis Distance-based Two-stage Robust State Estimation of Power System

针对在电力系统实际应用中,非高斯噪声会大大降低基于高斯测量噪声的常规状态估计方法的性能,提出一种非高斯噪声下基于马氏距离的两阶段鲁棒常规状态估计方法。首先,基于马氏距离计算得到多同步相量测量测点的固定最佳缓冲长度,建立基于常规监视控制和数据采集系统测量的极大似然估计器,将结果进一步与同步相量测量测点结合以在第二阶段中实现线性鲁棒估计。然后,在多种情况下基于IEEE-39节点系统的数值测试验证所研究方法的可行性,结果表明该方法可以约束坏数据和估计残差的影响,对39个节点的电压相角和电压模值的状态估计能力优于WLS估计算法10倍左右,且绝对估计误差始终稳定在一个更低的范围。

Aiming at the fact that non-Gaussian noise will cause great reduction in utility of conventional Gaussian measured noise-based state estimation methods in the practical application of power system,this paper proposes a two-stage robust conventional state estimation method based on Mahalanobis distance under non-Gaussian noise.First based on the Mahalanobis distance calculation,the fixed optimal buffer length of the multi-phasor measurement point is obtained,and a maximum likelihood estimator based on conventional monitoring control and data acquisition system measurement is established,and the results are further combined with the synchronizer measurement point to achieve linear robust estimation in the second stage.Then based on the numerical test of IEEE-39 node system in a variety of cases,proposed method is verified feasible and capable of limiting the influence of bad data and estimated residuals,as the state estimation ability of the proposed method for voltage phase angle and voltage modulus of 39 nodes is about 10 times better than the WLS estimation algorithm,along with an absolute estimation error maintained in a lower range.