Talk:Model predictive control

Systems Unassessed

	Systems science portal This article is within the scope of WikiProject Systems, which collaborates on articles related to systems and systems science.SystemsWikipedia:WikiProject SystemsTemplate:WikiProject SystemsSystems
???	This article has not yet received a rating on Wikipedia's content assessment scale.
???	This article has not yet received a rating on the project's importance scale.
	This article is not associated with a particular field. Fields are listed on the template page.

The article is at least misleading if not completely wrong. And furthermore is hardly understandable. Rewrite from someone with a clue about MPC needed.

Is this version any better ? Encyclops 00:11, 31 July 2006 (UTC)[reply]

Misleading statements

I think the statement:

"Despite the fact that most real processes are approximately linear within only a limitted operating window..."

is very misleading. Many, many real processes of practical importance, and those tackled using MPC are highly nonlinear (e.g. water control systems).

In fact, all processes are nonlinear. Another point is that the cost function is not explictly defined for a mpc. it depends on purpose and economical reasons. "The MPC then sends this set of independent variable moves to the corresponding regulatory controller setpoints to be implemented in the process." As far as I know only the first control move is applied to the process not the set. "linear MPC approaches are used in the majority of applications with the feedback mechanism of the MPC compensating for prediction errors due to structural mismatch between the model and the plant" Not only for structural mismatch but also for exogenous disturbances. — Preceding unsigned comment added by 87.78.92.134 (talk • contribs)

I agree, this needs work. There are two main reasons to use model-based predictive control. One is that there are multiple, interacting control inputs. Feedback alone won't handle that; some form of model that relates the inputs to the outputs is needed. That's what the article currently covers.

The second use of model-based predictive control is when the system has significant lag. If the delays in the system between a control change and and an output change are large, and there's some source of disturbance that affects the controlled value, feedback control can't keep up. The classic situation is heating for large buildings, where there's a long delay between requesting heat for an inside space and getting heat from a faraway heating plant. An outside temperature sensor is often used as a disturbance input, so that a drop in outside temperature cranks up the heating system in anticipation of heat demand as the building structure cools. Historically, such things are manually tuned, but there's a trend towards making the whole system adaptive. Here's a limited reference from Google's book scanning: [1].

This stuff can be explained better. It's really not that complicated. --John Nagle (talk) 04:10, 10 March 2008 (UTC)[reply]