Research On Distributed State Estimation Method Of Smart Distribution System

Smart distribution network is an important part of smart grid.To increase the penetration of distributed generation,and enhance the supply-demand interaction,distribution system must be equipped with the more flexible control method and more efficient optimization capability.All above is based on the observability,which means the state of the system is estimated with the data collected and the information detected.Distribution network usually has a large number of nodes and complex operating mechanism.Hence compared with the centralized state estimation method,the distributed state estimation method has advantages in the computing efficiency,communication load and protecting privacy of the agent.This thesis is concerned with the distributed state estimation method of distribution system according to the model,algorithm and application.Under the background of current measurement system,the measurement function is always nonlinear with the state variable,which makes the state estimation model based on the weighted least square method nonconvex.The distributed algorithms of the convex optimization problem cannot be applied to the state estimation.Constructing the state variable and the equivalent measurement by stage,a three-stage state estimation model is set up.In the three-stage state estimation model,the state estimation which are both convex is implemented in the first stage and the third stage and the nonlinear transformation is implemented during the second stage.For the different areas connected by the tie line in the distribution network,the coupling constraint of them is the consistency of the boundary state.The lagrangian relaxation method can be employed to relax the coupling constraint into the objective function and it can be combined with the gradient method to propose a distributed state estimation method.What’s more,with the augmented lagrangian method,the coupling constraint can be taken into the objective in a form of quadratic term.Another distributed state estimation method is proposed by combing the augmented lagrangian method and the altering direction of lagrangian multiplier.Case studies are provided analysis the efficiency of these two distributed methods.Considering the AC-DC hybrid distribution system is a trend in the future,the distributed state estimation method is applied into the hybrid system.The AC subsystem and DC subsystem can estimation their own states based on the local information independently.The results of case studies reveal the advantages of the decentralized algorithm.