An Online Non-Stationary Simulation Optimization Approach Based on Regime Switching
The work develops an online simulation optimization approach called RSOBSO to handle problems where environments change over time (non-stationary). It uses a Bayesian framework and a Markov Switching Model to manage prediction uncertainty from regime changes and input uncertainty from estimating unknown parameters. This method is solved using a metamodel that reuses past simulation results for fast, robust decision-making.