A distributed implementation of multi-area power system state estimation on a cluster of computers

Abstract

This paper presents an efficient weighted least squares (WLS) distributed algorithm for multi-area power system state estimation including measurements provided by the supervisory control and data acquisition (SCADA) system and the synchronized phasor measurement units (PMUs). Each area performs independently its own state estimation using local measurements and exchanges border information at a coordination level that computes the system-wide state. The basic criterion of partitioning a power system into several control areas is to have areas as equal in size as possible, so that the workload on each area processor is as balanced as possible, and interconnections between distinct areas be limited, as much as possible, to reduce the amount of inter-process communication necessary. A true distributed implementation on a cluster of computers running Linux is developed. The code is written in C and sparse linear algebra routines from the Portable, Extensible Toolkit for Scientific computation (PETSc) library are used. Communication among area processors and the coordinator processor is implemented by using the MPICH2 implementation of the message-passing interface (MPI) protocol. By using extensions MPICH-G2, ES-MPICH2, MVAPICH2 of MPICH2 protocol, a portable Web-based distributed multi-area state estimator can be developed based on geographically distributed and heterogeneous cluster of computers. The proposed distributed approach has the same accuracy and redundancy level as the centralized estimator and provides the same solution only from the strictly mathematical point-of-view, by assuming that the measurement arrival delays and skew errors are identical to those of the centralized state estimator. Numerical simulations with a 1180 bus system, partitioned into equally or unequally sized area networks, demonstrate the efficacy of the proposed implementation in comparison with a centralized WLS state estimation.