Active Distribution System State Estimation Using Photovoltaic Power Forecasting

With the increasing shortage of fossil energy and the pressure of environmental protection,the development and utilization of renewable energy has attracted more and more attention.In recent years,various distributed generations(DG)have been connected to the power grid continuously,which makes the power flow of distribution network change from one-way to two-way,and the change of topology structure is more frequent.Distribution System State Estimation(DSSE)provides data support for distribution management system,which is the guarantee of safe,stable and efficient operation of distribution system.Although the use of new electronic devices provides an opportunity to improve the observability of the system,there are still some problems on the distribution network side,such as low redundancy of real-time measurement,backward dispatching mode,insufficient absorption capacity of distributed energy and low automation level,which limit the high permeability of distributed energy.In this paper,the problem of inadequate real-time measurement for active distribution system is studied.The main work is divided into the following four aspects:1)A photovoltaic power forecasting model based on PCA-GA-ELM is proposed.Firstly,principal component analysis(PCA)is used to reduce the dimension of the input of the forecasting model to reduce the complexity of the prediction model and improve the operational efficiency.Secondly,genetic algorithm(GA)is used to optimize the initial weights and thresholds of the extreme learning machine(ELM)to improve the prediction accuracy and generalization ability of the forecasting model.Finally,the forecasting model is trained and predicted by using the actual data of photovoltaic power plants in Australia,and the forecasting effect of the model is verified.2)Pseudo-measurement error modeling.The photovoltaic power forecasting results are used as pseudo-measurement to improve the measurement redundancy of DSSE.In the process of DSSE,measurement weights have a vital impact on the accuracy of state estimation.Measurement weights are usually set according to the measurement accuracy of instruments.In this paper,forecasting errors are modeled by Gaussian Mixture Model(GMM),and the weighted standard deviation is used to set the weights of pseudo-measurements,which makes the results more reasonable.3)Active distribution system state estimation using photovoltaic power forecasting.Taking IEEE33-bus as an example,eight photovoltaic power stations are connected to the distribution system,and three schemes are designed to verify the improvement of the state estimation accuracy of active distribution system by the proposed pseudo-measurement modeling method.4)Detection and identification of bad data.Two cases are designed to verify the improvement of the proposed pseudo-measurement model and pseudo-measurement weight model for bad data detection and identification based on IEEE33-bus.The simulation results show that the proposed photovoltaic power forecasting model based on PCA-GA-ELM has high prediction accuracy and strong generalization ability.Meanwhile,the proposed pseudo-measurement model provide more accurate pseudo-measurement values for DSSE,and set the pseudo-measurement weight reasonably,which improves the estimation accuracy of DSSE.Moreover,the proposed pseudo-measurement model improves the ability of detection and identification of bad data,making the measurement with gross errors from undetectable to detectable and from detectable to detectable.In conclusion,the pseudo-measurement modeling method proposed in this paper has practical engineering application value for DSSE.