Research On Distribution Power System State Estimation Under Multi-source Information

With the development of distributed generation technology,the DG(distributed generation)penetration is becoming higher in distribution power system.The RES(renewable energy system)which is a kind of intermittent power generation source occupies a big part of DGs.As the output of RES units has great uncertainty,the power flows in distribution system become more changeable and uncertain,thus making the operation process more complex.In order to make the most use of RES and the already existed assets in the distribution system,a better functioned method to predict,monitor and percept the operation state information is in demand.That is,a distribution system state estimation method which adapts well to the new features of active distribution system is needed.The research work tries to solve two problems:1.Increase the state estimation frequency to provide more timely system operation state information;2.As the DGs in distribution system is making a great influence to the system operation.We try to reduce the influence to state estimation results by analyzing the operation modes of DGs.The thesis analyses the feature of current distribution system,including the structure characteristics and the measurement-communication system.Based on these characteristics,the thesis discusses to utilize multi-source information-information provided by SCADA(Supervisory Control and Data Acquisition)system,quasi real-time information provided by the AMI(Advanced Measurement Infrastructure)based on smart meters and the newly researched micro synchronized phasor measurement units(micro-PMU,or ? PMU)information-to conduct distribution system state estimation.The thesis firstly summaries the basic procedure of distribution system state estimation and analyze the advantages of the branch current based method based on the structure and parameter features of distribution power system.This part also illustrates how to make use of the structure variables to calculate the relationship between different electric variables in the distribution system.A branch current reduction method is used to avoid the calculation problems in dealing with the zero-injection power information.A comprehensive branch current based distribution system state estimation method which combines the FASE(Forecasting Aided State Estimation)method and the WLS(Weighted Least Squares)method is proposed.This method applies to the case when the real-time measurement devises is comparatively adequate.By comparing and analysis of the prediction information and the field measured information,the judgment of if any abnormal situation occurs is conducted.The proposed method can also discern the situations between bad data invasion and the sudden change of operation.Simulation is conducted to prove the effectiveness of the proposed approach.GMM(Gaussian Mixture Model)is utilized to conduct probability characteristic analysis of DG output,approximating the probability density function distribution by a linear combination of a confined number of Gaussian distribution.Each Gaussian component or an equivalent one of several Gaussian components represents one operating pattern of DG,where the mean value represents the cluster center and the variance represents the uncertainty.As the DG output recorded by AMI system update comparatively slowly,the operating pattern may be changed during two updating moments and for this reason the operating pattern analysis of DG is needed.When the operating pattern of a DG is proved to be changed,the operating pattern information is used to make the pseudo measurement for state estimation calculation.Simulation results show that under a certain real-time measurements configuration,this approach can improve the calculation accuracy and better reflect the operating condition change.