Forecasting Aided State Estimation Considering State Time And Spatial Correlations

Power system state estimation plays a vital role for reliable and secure power system operations.Power system state estimation is able to calculate the unmeasured states,the results will be sent to the Energy Management System application functions such as the contingency analysis,load forecasting and optimal power flow.However,traditional static state estimation lacks forecasting ability and no realistic prediction of the states can be obtained.This deficiency motivated the researchers to propose forecasting aided state estimation to provide predicted state vector at the next time instant.Forecasting aided state estimation uses the present state of the power system along with the knowledge of the system’s physical model to predict the state vector for the next time instant.Once the measurements at the next instant arrive,the predicted state vector is filtered to obtain an optimized estimate.Forecasting aided state estimation helps to detect,identify and reject bad data and hence improves the estimation performance.Forecasting aided state estimation algorithms proposed so far have considered state transition matrix as diagonal.This means that the power system states are assumed to vary independently from each other.This assumption is not realistic in power systems in the presence of microgrids,distributed and intermittent power generation,which are connected to the grid sometimes through radial buses.The grid of future is becoming more decentralized resulting in the decrease of number of buses with large amount of power injections.Therefore,these electrical hubs lose their significance and dominance in the grid and the spatial correlation among the states throughout the grid become more observable.Taking spatial correlation into account also provides the operator to be able to restore observability when a state is unobservable due to loss of telemetry.So,this paper proposes a novel model which takes the time correlation and spatial correlation into account to determine state transition model for forecasting aided state estimation,the main work are summarized as follows:(1)Firstly,this paper gives a brief review of time-series analysis.Several common used methods are introduced.Moreover,the techniques to identify and fit a proper model for a set of time-series data are compared by simulations.(2)In terms of state forecasting considering time correlation and spatial correlations,a sparse,non-diagonal and time-variant state transition matrix is proposed.The main work is using autoregressive model to model the time correlation in power system,while the spatial correlation among the system states is modeled using vector autoregressive model.Then,the performance of the proposed technique for systems under normal operating states and under loss of observability are compared with the common used exponential smoothing methods by simulations.(3)In terms of forecasting aided state estimation considering time correlation and spatial correlations,a fixed AR/VAR model is used to forecast the state of the buses.The main work is to compare the estimation results using the proposed technique and Silva state forecasting technique for systems under normal operating states and under loss of observability,the conclusion of the results are followed.