Study On Distribution System State Estimation

The integration of distributed generations and emerging loads like electric vehicles brings more uncertainty into distribution system and makes distribution system operate in a much more complicated and changing state.Highly accurate and real-time sensing of distribution system should be achieved to guarantee secure and stable operation,and state estimation is one of the most effective tools to get the system state.Therefore,this dissertation focuses on distribution system state estimation.The contributions are summarized as follows:1)A new real/reactive power decoupled three-phase numerical observability analysis method and a new critical data identification method for distribution systems are proposed.Strict theoretical proofs of the decoupled and non-iterative performance of the methods are provided.The methods can efficiently identify all unobservable branches and critical data in a non-iterative manner and the computation burden is greatly reduced.Furthermore,since all entries in the Jacobian matrix are non-negative integers,the proposed methods present good numerical performance.2)A novel three-phase measurement placement model and a bad data identification method for distribution system are proposed.Firstly,measurement placement problem is modeled considering both network observability and state estimation accuracy,aiming to maintain the network observability against loss of any measurement and minimize the total variance of state estimation error.Secondly,an zone modification based three-phase bad data identification method is proposed,which devides the network into several zones via critical measurements.Each time a bad data is identified,variables in the same zone of that bad data rather than the whole system of that phase need to be updated.Hence,high dimensional matrix computation is avoided and the computing speed is greatly improved with acceptable estimation accuracy.3)An analytical feeder remote terminal unit(FRTU)deployment approach for the reliability improvement of distribution systems is proposed.The FRTU deployment problem is modelled as an integer programming problem,which aims to minimise the total cost which includes the life cycle cost of FRTUs in addition to customer interruption cost with the reliability index of average service available index as the constraint.Both overhead and underground distribution feeders are considered.The reliability and economic indexes are analytical expressed by decision variables so that the model can be solved by large-scale commercial solvers in an efficient manner.4)A guaranteed two-stage state estimation for three-phase unbalanced distribution systems is proposed to obtain the upper and lower bounds where the true state is guaranteed to be contained.The first phase is to calculate the reliable distribution power flow with all pseudo measurements of low accuracy.The affine arithmetic based threephase back/forward sweep distribution load flow method is adopted.The calculated voltage intervals are then used as input of the second phase,where real-time measurement with higher accuracy are used as more strict constraints to further contract the initial boxes.In this phase,concave and convex envelope technique is employed to build the linear relaxation model which can be efficiently solved.