Model predictive control (MPC)
also: MPC · receding horizon control
A control method that repeatedly optimizes actions over a short future horizon using a model, respecting limits while minimizing a cost such as energy or error.
At each control step, MPC uses a model to predict how the system will respond to different future actions, chooses the sequence that best meets the goal (track a target, save energy, reduce wear) without violating constraints (actuator limits, safe temperatures, comfort bands), applies only the first action, then repeats with fresh measurements.
Its strengths are explicit handling of constraints and the ability to use forecasts such as weather, electricity prices, occupancy or a planned path. That makes it common in process plants, building HVAC and energy management, battery systems and vehicles.
Where it bites: it needs a reasonably accurate model, more computation than PID, and careful engineering so the optimization always returns a usable answer in time. For a simple single loop it is usually overkill.
Founder takeaway
Consider MPC when constraints, forecasts or energy cost are central to the product's value; otherwise start with simpler loops.
engineer review pending