Research On Dynamic State Estimation Method Of Distribution Network Based On Koopman Framewor

Distributed energy resources(DERs),demand-response loads,and measuring devices with different data rates are changing the character of the distribution system.Changes in load composition,especially the increasing popularity of fluctuations DERs,make distribution system state nodes change more frequently and quickly,so it is very important to accurately track and estimate distribution system state.In addition,the study of the state estimation of distribution system is necessary for the realization of protection,optimization and control technologies and various other functions envisaged by the smart grid.To address the problem that it is difficult to build an accurate dynamic state estimation model for distribution system due to the complexity and dynamics of distribution systems,this paper invokes Koopman operator theory to achieve linearization of distribution system in a non-approximate case.In a data-driven manner,a high-dimensional linear state space expression for the distribution system model is constructed,and this process only requires input and output data without any a priori model knowledge.And a series of researches are conducted on the dynamic state estimation algorithm of the distribution system based on Koopman framework,the main works are as follows:1.Considering the complexity and dynamics of the distribution system,the state estimation model of the distribution system is linearly represented in a data-driven way based on the theoretical framework of Koopman operators,and the state transfer matrix and the measurement equations in the DSE model of the distribution system are replaced by the Koopman observation framework.A novel Koopman Kalman filter(KKF)-based dynamic state estimation algorithm is applied to the distribution system and combined with H∞ filtering to reduce the interference of uncertainties such as distribution system load fluctuations on the state estimation results.2.To address the problem that the accuracy of Kalman filtering decreases in the non-Gaussian distributed measurement noise environment,we combine KKF with particle filtering and propose a dynamic state estimation method based on Koopman Kalman particle filtering(KKPF),so that the Kalman filter based on the theoretical framework of Koopman operator can be applied to the non-Gaussian distributed measurement noise environment.The simulation results demonstrate that the proposed method can be used in both Gaussian and non-Gaussian distributed measurement noise environments.3.The performance of the state estimation method is degraded by the measurement outliers,so a distribution system dynamic estimation method based on the innovation-saturated Koopman Kalman filter(IS-KKF)is proposed.By introducing an innovation-saturated mechanism to limit the measurement outliers,the outliers are not allowed to interfere with the process of the filtering step in Kalman filtering.The mechanism is simple in structure,computationally efficient,and does not require measurement redundancy,which makes it suitable for real-time applications of distribution system dynamic state estimation.