Distributed Moving Horizon Estimation for power systems

We adapt and apply a known algorithm for Distributed Moving Horizon Estimation (DMHE) to power systems. In this distributed approach, the power system is partitioned into several control areas. At each time step the state of the whole system is estimated locally in each area, by solving a local optimization problem. A consensus weights update step is performed in which the covariance matrix of the initial penalty is dynamically computed to ensure convergence of DMHE algorithm. DMHE converges to the centralized solution of moving horizon estimation (MHE) within a reasonable number of time steps. Numerical simulations with the IEEE 14-bus system and experiments on a lab microgrid show better results than those obtained from the standard approach using weighted least squares (WLS), if known constraints on states and noise are exploited.