Modeling

I recently encountered one process control topic "Model Predictive Control or simply MPC" as most of my time is to do research on modeling and simulation. Since it was not taught in any of my control systems courses, I am quite fascinated to know more about this.

The underlying concept comes from human intuition. As most of control theories come from real implementations such as PID, MPC also falls into the same basket.

The below image tells very nicely about MPC idea. (Image taken from Model Predictive Control, by E. F Camacho, Carlos Bordons )


It is very simple. Suppose, if you are driving a car it is very obvious that you know the path at least couple of meters ahead of you. That information is well enough you to how to drive your vehicle. In mathematical terms, the future control action needs the prediction of the target path (or prediction of the future errors), in contrast to PI control, where the control action needs past errors.

Of course I am not an expert to explain more about this, but quite interested to show off my interest in this field. Majority of industrial process use this theory in some manner and DMC (Dynamic matrix control, by Cutler, Ramakar, 1980) is the widely used Model predictive control.