Research On Distributed State Estimation Of Distribution Network

As one of the vital functions of the EMS,state estimation uses the measured data obtained from the Supervisory Control And Data Acquisition(SCADA)system to analyze the network topology,establish the network model,and calculate the actual running state of the system through the optimal estimation criteria,so as to provide a complete and reliable database for further analysis and control of the system.Therefore,the safety and economic operation level of power system are directly affected by the accuracy and real-time performance of power system state estimation.The state estimation matrix scale expands unceasingly with the expansion of power system scale.In addition,state estimation need to be repeated when optimizing measurement configuration or proceeding static security analysis,which makes the computation speed of traditional centralized state estimation cannot meet the requirement of the real-time analysis and control of power grids.This thesis presents an automatic partitioning method for distributed state estimation of distribution network.Considering calculation quantity equilibrium of each partition,the method realizes the automatic partitioning of the distribution network by virtue of the rapid search capability of the network in the graph theory,so as to realize parallel computing of state estimates.The dimensionality of the state estimation is reduced by partition,so as to reduce the computation scale and speed up the computation.At the same time,the local bad data can be limited to the current sub-region,so as to avoid the local bad data affecting the estimation accuracy of the whole network,or causing the non-convergence of the whole network estimation.The method of correcting the weight coefficient is adopted in this thesis to enhance the robustness of the state estimation algorithm.This method introduces exponential function into the weight coefficient,which uses exponential variable weight coefficient to reduce the weight value of the bad data in the iterative process,thus reducing the adverse influence on the target function and estimated results.Finally,based on the IEEE-118 nodes system,the program is written and improved in the FORTRAN6.5 compile environment.The distribution network ispartitioning by the automatic partitioning method,and calculated by distributed state estimation.The result verifies the proposed method can achieve rational partition of distribution network,improve the computational efficiency of state estimation.Meanwhile,the robust state estimation which adopts modified weight coefficient,can minimize the impact of bad measurement data.