Fast LAV Estimation via Composite Optimization

Accurate and robust power system state estimation (PSSE) is an essential prerequisite for reliable operation of smart power grids. In contrast to the commonly employed weighted least squares (WLS) one, the least-absolute-value (LAV) estimator is well documented for its robustness. Due to the non-convexity and non-smoothness however, existing LAV implementations are typically slow, thus inadequate for real-time system monitoring. In this context, this paper puts forward a novel LAV estimator leveraging recent algorithmic advances in composite optimization. Concretely, the estimator is based on a proximal linear procedure that deals with a sequence of convex quadratic problems, each efficiently solvable by means of either standard convex optimization methods, or the alternating direction method of multipliers. Simulated tests using two IEEE benchmark networks showcase its improved robustness and computational efficiency relative to several competing alternatives.