State Estimation for Large-Scale Systems Based on Reduced-Order Error-Covariance Propagation

We compare several reduced-order Kalman filters for discrete-time LTI systems based on reduced-order error-covariance propagation. These filters use combinations of balanced model truncation and complementary steady-state covariance compensation. After describing each method, we compare their performance through numerical studies using a compartmental model example. These methods are aimed at large-scale data-assimilation problems where reducing computational complexity is critical.