Research On Distribution Network State Estimation Based On Unscented Particle Filter

The distribution network is an important part of the power system,and its safe and stable operation is closely related to the safety and stability of the power system.Distribution network state estimation is the core of the distribution management system and the basis and premise for ensuring the safe and reliable operation of the distribution network.Because the original measurement data provided by the distribution network data measurement collection and transmission system has various random noise interferences.If it is directly used to analyze the operation of the distribution network,it will cause erroneous judgment,affect the stability of the power grid and the safety of operation,so the measurement data must be filtered,that is,the state estimation process.Based on a large amount of reading and research of domestic and foreign literature,this article mainly carried out certain research work from the following aspects:Firstly,in-depth research and analysis of the state estimation method based on the Kalman filter framework and the state estimation method of the particle filter theory,comparative analysis of the advantages and disadvantages of the two types of methods.In view of the characteristics of the three-phase imbalance of the distribution network,three-phase state estimation is required.Based on existing literature,a dynamic space model of the distribution network state estimation is established,that is,a state transition model and a mixed measurement model based on Holt twoparameter exponential smoothing.Secondly,in view of the shortcomings of the unscented Kalman filter(UKF)method and the basic particle filter method,the unscented particle filter(UPF)state estimation method is introduced.This method first uses the feature of high accuracy of UKF estimation to generate importance density function for particle filtering to transfer the predicted particles generated by sampling to the high likelihood region of filtering,and then uses particle filtering theory to realize state estimation,which improves the state estimation performance.the UPF method is applied to the distribution network state estimation to demonstrate the effectiveness of the proposed method.At the same time,the estimation error statistics and fitting tools are introduced into the simulation analysis and the overall performance measurement value of the estimation error is calculated.The state estimation performance of the proposed algorithm is qualitatively and quantitatively analyzed.Thirdly,in view of the defect that the quality of the density function generated by the unscented particle filter is not high and the robustness is not strong,which leads to the defect that the UPF state estimation algorithm still has limited filtering accuracy and is not robust,an improved unscented particle filter is proposed.The improved algorithm includes two aspects: one is to control the distribution of Sigma sampling points by adaptively adjusting the scale correction factor of the unscented transformation to improve the quality of the importance density function(recommended distribution);the other is to replace the traditional UKF with strong tracking unscented Kalman filtering.When the system measurement is abnormal,the Kalman filter gain is adjusted online by introducing a fading factor,thereby enhancing the ability of the UPF filtering algorithm to deal with abnormal conditions,that is,improving the state estimation algorithm robustness.