Application of partitioned-based moving horizon estimation in power system state estimation

Abstract

The Partitioned-based Moving Horizon Estimation (PMHE), developed previously by others, is applied to the power system state estimation problem in this paper. The constraints on state variables and noises are taken into account in this scheme. In this distributed approach, the network is partitioned into several non-overlapping and observable areas. The global Jacobian matrix is required during the initial time before approaching the converged states. Only the estimated information data between neighboring areas are exchanged afterwards. The communication traffic is thus significantly reduced compared to a centralized solution. Meanwhile, each area estimates its local states by solving a smaller size optimization problem. The optimization problem is, therefore, scalable. PMHE converges to the centralized solution of moving horizon estimation (MHE) within finite time steps. Numerical simulation with the IEEE 14-bus system shows the convergence of PMHE. Further, the estimated states are better than those from the weighted least squares (WLS) with large outliers.