Model predictive Control 101

What is Model Predictive Control?

Pallav Rawat
3 min readMay 6, 2021

If you are wondering what is model predictive control, this diagram would give you a fair clue:

The Predicted Output tries to clone the Reference Trajectory at every instant of time.

Model Predictive Control Scheme

The way a human driver steers a car is itself an example of MPC. The reference trajectory is the lane to be driven in, and the effects/countereffects on steering wheel angle is the model based approach for tracking the reference.(example : https://hortovanyi.wordpress.com/2017/06/06/model-predictive-controller-project/)

Uses:
First used in 60’s
Used in process industries and chemical plants since 80's
Now has become a hot topic and started to be used in:
(1) Automotive industry
(2) Aerospace Industry
(3) Smart Electrical Grids and Power systems
and many other industries

Some Pros and Cons:
§Pros: Extremely flexible control design approach:
— Prediction model can be multivariable, w/delays, time-varying, w/disturbances, …
— Can exploit available preview on future references and measured disturbances
— Handles constraints on inputs and outputs
— (Auto)tuning similar to Linear…

--

--

Pallav Rawat
Pallav Rawat

Written by Pallav Rawat

Data. Anti-fragility. DIY. Hacks.

No responses yet