Multiarea state estimation with legacy and synchronized measurements

Abstract

In this paper, we propose a multiarea state estimator based on successive convex approximation technique. Our scheme is hybrid since it exploits both non-linear and linear measurements coming from legacy meters and Phasor Measurement Units (PMUs), respectively. The resulting nonconvex optimization problem is solved in an iterative manner. In each iteration, the nonconvex terms in the cost function are approximated by a strongly-convex function. The resulting convex problem is distributedly solved by means of the so-called Alternating Direction Method of Multipliers (ADMM). We also prove convergence to a stationary solution of the original nonconvex state estimation problem. Performance is numerically assessed for the IEEE 14bus and 30-bus test cases. Other optimization schemes from the literature are used as a benchmark.

5G Mobile Cellular Networks: Enabling Distributed State Estimation for Smart Grids

“…Here, the SE problem is casted into a semidefinite programming framework and solved via convex semidefinite relaxation techniques, both in centralized and decentralized settings. In [10], the authors propose a hybrid multi-area state estimator based on successive convex approximation (SCA) and ADMM. The proposed distributed approach is equivalent to the centralized case in terms of estimation accuracy and is able to operate in broader scenarios where the semidefinite relaxation approach fails.…”

“…In distributed approaches [8]- [10], on the contrary, each local processor communicates only with its neighbors, since no central authority exists. The authors in [8] propose a distributed SE scheme based on primal-dual decomposition.…”

“…To decentralize an optimization problem, ADMM decouples the objective function with consensus variables. Consensus variables introduce equality constraints into the optimization problem, separating it into a number of subproblems [9], [10]. The resulting ADMM algorithm can be interpreted as an iterative message passing procedure, in which the agents solving the subproblems (e.g., utilities in the multi-area SE problem) exchange consensus and dual variables until convergence.…”

“…The BP algorithm is implemented as a BP-based distributed Gauss-Newton method described in [13], which can be interpreted as a fully distributed Gauss-Newton method. The ADMM-based algorithm is based on [10], where the SCA scheme in the outer loop is combined in a distributed fashion within the iterative framework of ADMM, that constitutes the inner loop. To evaluate both algorithms, we use the Root Mean Square Error (RMSE) after each iteration k (RMSE k ), normalized by RMSE of the centralized SE algorithm using the Gauss-Newton method after 12 iterations (RMSE GN ).…”

On the impact of LTE RACH reliability on state estimation in wide-area monitoring systems

“…Traditionally, problem (2) is solved via gradient-based schemes [6]. Other efficient solutions based on convex relaxations [7], [8] and graphical models have been also proposed during the latest years [9]. As discussed before, a crucial element in order to improve the monitoring quality of the system is the phasor technology.…”