Dynamic State Estimation of Cyber-Physical Power and Energy Systems Based on Bad Data Identification

Abstract

Dynamic state estimation is an important approach to realize power network state perception in Cyber-Physical power and energy systems. However, more and more bad data are transferred into power grid energy management system because of data attack on transmission line or damage of measuring equipment. Meanwhile, as clean energy access to the power grid, the random nature of these sources leads to a large amount of fluctuating measurement data in the system, which is true data. The traditional robust Cubature Kalman Filter (CKF) method cannot distinguish bad data from fluctuating data. Fluctuating data are regarded as bad data, and the real characteristics of the data are lost. Based on the above deficiencies, a bad data pre-identification method based on big data is proposed, and improved robust volume CKF is conducted on each node of the Cyber-Physical power and energy systems. The innovation covariance based on the innovation sequence is calculated firstly and the measured variance with the total variance is compared to obtain the first-level identification; Secondly the correlation of the data according to the measurement sequence is calculated, and the secondary identification is conducted to distinguish the fluctuation data from the bad data. Simulation results show that the improved method can reduce the impact of bad data, while ensuring the authenticity of the measurement data, and improving the accuracy of state perception.