State estimation for large-scale partitioned systems: a moving horizon approach

In this paper we propose novel state-estimation methods for large-scale discrete-time constrained linear systems that are partitioned, i.e. made by coupled subsystems with non-overlapping states. We focus on moving horizon estimation (MHE) schemes due to their capability of exploiting physical constraints on states and noise in the estimation process. We propose three different partition-based MHE (PMHE) algorithms where each subsystem solves reduced-order MHE problems to estimate its own state. Different estimators have different computational complexity, accuracy and transmission requirements among subsystems. Numerical simulations demonstrate the viability of our approach.