Two-Scale Model Predictive Control for Resource Optimization Problems with Switched Decisions

Model predictive control (MPC) is widely used in resource optimization problems because it naturally deals with bounded controls and states allows information to be included. However, at each sampling instant, an problem must solved. Resource switching actions lead integer decision variables, which are computationally costly, particularly when the number of variables large. As a result, approach directly discretizing (DD) derive mixed-integer linear program (MILP) sets fundamental limitations on MPC rate owing computational time required solve problem. In this paper, we propose two-scale algorithm (TSOA) for MPC. On first-scale, entire prediction horizon considered provides optimal resources interval constant weighting cost. This may cast as (LP); thus, tractable even large constraints. second-scale, nature variable recovered by posing MILP deploy computed previous scale. manner, solved shorter than horizon, thus reducing The simulation results demonstrate advantages proposed compared direct discretization optimization.