State Estimation Of Distribution System Based On Adaptive Robust Unscented Kalman Filter

Distribution system state estimation is the core function of the monitoring and dispatching center of the distribution system.The purpose of state estimation is to obtain the real-time and accurate operating status of the distribution system based on the system network structure and measurement data,in order to regulate the operation of the distribution system,prevent the occurrence of safety accidents,and ensure the safe,stable and economic operation of the system.Currently more widely used is static state estimation based on traditional weighted least squares method.This method can only reflect the system state of a single time section,and cannot display the data characteristics of dynamic multiple time sections.The emergence of Kalman filter can meet the requirements of dynamic system state estimation,and occupies an important position in the prediction of system operation trend.In this thesis,the basic principle of state estimation for distribution system is introduced,and an improved Kalman filter algorithm is used for dynamic modeling.The unscented Kalman filter algorithm and the extended Kalman filter algorithm are both non-linear filtering methods,when dealing with high-dimensional non-linear distribution system,the unscented Kalman filter has better filtering effect.The unscented Kalman filtering algorithm used in this thesis cannot accurately reflect the dynamic system noise due to the use of fixed constants when faced with system noise,and cannot effectively counteract the effect of measurement gross error on filtering when gross measurement errors occur.Therefore,a new projection statistics-adaptive robust unscented Kalman filter algorithm is proposed to improve the stability of the system during the estimation process.By adding a Sage-Husa time-varying noise filter and a robustness estimator based on projection statistics,the system’s adaptability in the case of unknown noise and the filter processing performance in the case of measuring gross errors are improved.In order to further improve the filtering accuracy,the hybrid measurement is also dynamically modeled in this thesis.The curve synchronization is used to ensure the synchronization of PMU data and SCADA data at the sampling frequency,and the conversion model is used to integrate two different types of measurement data to improve the accuracy of the measurement data.The various algorithms mentioned in this thesis are modeled in MATLAB and combined with IEEE 3 3-node and PG&E 69-node systems for simulation testing.Among them,the adaptive measurement unscented Kalman filter algorithm with hybrid measurement has achieved good filtering effect in the test system containing distributed generation sources,and the effectiveness of the algorithm is verified through comparative experiments.