Robust Distributed State Estimation For Active Distribution Networks

With the high penetration of distributed generations(DG)and controllable loads(CL),the distribution network gradually changed from a passive form to active distribution networks(ADN).Compared with passive distribution networks,the core characteristic of ADN is that controllable resources can be controlled and utilized actively to realize active regulation and control of distribution operation,which is beneficial to improve the reliability and power quality of ADN.A good sensing ability for the real-time operation state of ADN is the premise to realize the active regulation of the distribution operating states where the measurement information is used to estimate the states of distribution networks.Distribution system state estimation(DSSE)is an effective way to realize situational awareness of distribution networks.ADN has the characteristics of large system scale,three-phase unbalance,high real-time data and accuracy requirements.Traditional centralized state estimation faces with great challenges in calculation accuracy and time.Thus,the development of distributed state estimation(DSE)has become inevitable.The core idea of DSE is to divide large-scale power grid into several small-scale sub-regions.Each sub-region calculate local state estimation in parallel or serial computing architecture.Then,information is exchanged between adjacent sub-regions or between sub-region and control center according to different requirements until all the convergence criteria are satisfied,which means the overall state estimation of ADN is achieved.In ADN,there are many factors affecting the performance of DSE.In this paper,we mainly focus on three aspects,network partition of ADN,PMU placement and communication failures.The main research works in this paper are as follows.(1)A network partition approach for DSE in ADN is proposed based on the topology analysis method and the PMU installation location,according to partition criteria of the equilibrium of node number and measurement redundancy in each sub-region,observability and connectivity of sub-regions.Firstly,the partition criteria are given according to the characteristics of DSE calculation.Network partition approach is established based on topology analysis approach including branch-line layer method and postorder-traversal algorithm,which has two stages,node numbering and network partition by topology searching to realize network partition.Sub-region decoupling is implemented at the overlapping nodes according to the equivalent relationships of the distribution network and the PMU measurements.(2)Multi-objective model for PMU placement is proposed considering economic,partition equalization and performance of DSE.Combined NSGAII-TS algorithm is used to solve this problem,which gives the Pareto solution set of PMU placement.Firstly,the interaction and coupling relationship between PMU placement,network partition and DSE are analyzed.According to the analysis results,the three objectives of placing PMU are extracted,which are economic,partition equalization and DSE performance.The multi-objective NSGAII-TS algorithm is used in PMU placement model to give the placement results.(3)A novel distributed state estimation with consensus coordination(CC-DSE)is proposed to deal with communication failures which has great impact on the performance of DSE.First,according to the partition and PMU placement method,the network is partitioned into several sub-regions.Then.DSE of ADN is implemented considering communication failures.The improved WLS algorithm with adaptive weights is proposed in local state estimation which is robust to inner-region communication failures.Consensus-based coordination is embedded into iterations of DSE to calculate inter-region exchanged data to improve the computational performance of DSE in terms of accuracy and efficiency.