Topologically-Partitioned Power System State Estination

Estimating the state of a large electric power system is an ill-conditioned computational task simply because of its size. This paper describes a technique which reduces the large state estimation problem to a set of smaller least-squares problems plus the solution of a square system. An observability algorithm developed earlier is used to partition the system into sub- networks whose states are estimated separately. These separate estimates are reconciled using the solution of an interconnecting subnetwork consisting entirely of measurements whose removal would render the full system unobservable. Under appropriate conditions, this partitioned state estimate is identical with the usual full least-squares estimate. Details of the implementation of this algorithm are given, and its application to the IEEE 30-bus network is described.