Model Predictive Control, MPC
Updated 2026-08-13
INTRODUCTION
English translation pending.
CORE DEFINITION
Model predictive control uses a model of the system to predict its behaviour over a finite future horizon, solves an optimization problem that minimizes a cost subject to constraints, applies only the first control move, and then repeats the whole computation at the next time step. The core proposition is that rolling re-optimization with feedback handles constraints and disturbances better than a fixed open-loop plan. The key qualification is that longer horizons accumulate model error and require more computation, and the optimizer must finish within the control interval to be usable.
SCAFFOLDING EFFECT
Reduce cognitive load
- Horizon choice: pick a prediction horizon long enough to see the consequences but short enough to compute. - Constraint list: state the hard limits the solution must respect before optimizing anything. - First-step rule: commit only the first move, then re-plan with fresh measurements.
Anchor fast decisions
Optimizing over a finite horizon produces a plan that respects the constraints, but only the first step is executed because later steps rest on predictions that will be stale. Re-solving with updated measurements corrects for the model error that has accumulated. The feedback loop is what gives the method robustness, since each cycle replaces prediction with observation.
MINIMUM ACTION
In progress 0/1Practice this model in one real situation:
account_treeGenealogyexpand_more
menu_bookReferencesexpand_more
Source support: Explicit
- en.wikipedia.orghttps://en.wikipedia.org/wiki/Model_predictive_controlverified
PRIVATE NOTES · Only visible to you
SAVED Q&A
ENTRY Q&A · Private saving available
Ask with a clear boundary
thinkingmodels answers from published entry context only.
Your question is sent to thinkingmodels. The answer uses public entry context only.
RELATED MODELS