Research Of Distribution System State Estimation

With power system quickly developing, Smart Grid has become an inevitable trend of development and reached a consensus in the world. The automation level of distribution network, which connects the power grid and the users as an important bridge, is one of the important index of Smart Grid. As far as I am concerned, the lack of real-time measurements in the distribution network can't obtain the reliable and high-precision dates. Therefore, it has seriously limited the automation level. Based on analyzing for the traditional state estimation methods in the thesis, this thesis does methodological and experimental research and states both static state estimation method and dynamic state estimation method. The main content of the thesis as follows:(1) Considering the static state estimation method for the distribution network, a power-transformation-based state estimation method for distribution network is developed. And the artificial neural network(ANN) and the Gaussian mixture model(GMM) are used to obtain pseudo measurements and its weight. Therefore, this method considerably improves the accuracy of computation.(2) An unscented Kalman filter is used for dynamic state estimation method of distribution network. Predictions are generated from an ultra-short term load forecasting using Holt-Winter model. Then initial dates and predictions based on cubic spline interpolation are used to obtain the pseudo measurements. The pseudo measurements are used in state estimation to realize the real-time tracking of state prediction.(3) All experiments of proposed two methods have been implemented in the standard distribution network model on MATLAB. The validity of the methods and conclusions is proved.