Research On Distribution Network State Estimation Method Based On PMU/SCADA Measurement Data

Distribution network operation is becoming more and more complex.The accuracy of state estimation results is an important guarantee for the safe and stable operation of distribution network.Phasor Measurement Unit(PMU)has been deployed in distribution network,which provides data basis for distribution network state estimation.However,due to the constraints of economy and complexity of technical transformation,it is impossible to achieve flill coverage of distribution network PMU in a short time.Therefore,PMU measurement and Supervisory Control and Data Acquisition(SCADA)measurement will constitute a reliable data source for distribution network state estimation in a long time to come.Therefore,it is of great significance to study the distribution network state estimation method based on PMU/SCADA mixed measurement data.In recent years,compared with static state estimation,dynamic state estimation is more in line with the dynamic characteristics of power system and does not need iteration,so it has become a research hotspot.The most commonly used method of it is dynamic state estimation based on Unscented Kalman Filter(UKF).However,the current UKF-based dynamic state estimation methods may be affected by unknown and time-varying process noise,resulting in degradation of filtering performance or even filtering divergence.Therefore,a dynamic state estimation method for distribution network based on Improved Robust Adaptive Unscented Kalman Filter(IRAUKF)is proposed in this thesis.IRAUKF uses a biased and unbiased estimator to construct noise statistic estimator.It can estimate the statistical parameters of process noise in real time in the process of state estimation.On the basis of guaranteeing the robustness of the algorithm,it reduces the loss of correction information and enhances the adaptability to time-varying process noise parameters in dynamic state estimation.In addition,a fusion estimation strategy of measurement information at different time scales is proposed based on IRAUKF.The state of the system can be updated at the time of PMU sampling,so that the latest state of the system can be mastered by using the high-frequency sampling characteristics of PMU.As the input of the state estimator,the line parameters are the necessary information to construct the measurement equation.The error of the line parameters will have a long-term negative impact on the state estimation accuracy.Therefore,identification and correction of erroneous parameters are of great significance to state estimation.In this thesis,a rolling correction method for error line parameters based on IRAUKF augmented dynamic state estimation is proposed.The regularized residual of line power flow measurement is obtained by using the results of dynamic state estimation,and the error parameters are identified based on the average of regularized residual at multiple times.Then,the identified error parameters are used as state variables to establish an augmented state space model.Aiming at the problem of insufficient redundancy of single measurement section,an augmented state space model of multi-measurement section is established.This method takes advantage of IRAUKF which can estimate process noise statistical parameters of parameter variables in real tine to realize rolling correction of error parameters.The simulation results show that the method can accurately identify and estimate parameters errors of single and multiple branches.In order to adapt to the rapid change of system state in some scenarios,it is necessary to calculate the system state quickly.In this paper,a fast state estimation method based on AE-ELM pseudo-measurement modeling is proposed by using Autoencoder(AE)and Extreme Learning Machine(ELM)on the premise that the dispatching center has accumulated abundant operation data.Firstly,the dimension reduction feature data of SCADA injection power measurement is obtained by AE,which is used as the input of ELM,and the real and imaginary parts of voltage are used as the output.The pseudo measurement model is obtained by training historical data.Then the pseudo-measurements are combined with PMU real-time measurements for fast linear state estimation to obtain the final state estimation results.The simulation results show that the method can improve the computational efficiency on the basis of guaranteeing the accuracy of state estimation to meet the needs of fast state estimation.