Adaptive Control of Nonlinear Stochastic Systems by Particle Filtering
Aaron Greenfield, Anthony Brockwell · 2003
We introduce an adaptive moving horizon control scheme for nonlinear stochastic systems. The scheme uses the recently developed particle filter to track the hidden state, as well as to estimate unknown parameters. In addition, expected costs are approximated by Monte Carlo integration where necessary. Although computationally intensive, the scheme has wide applicability, and we demonstrate its robustness in simulations.