classical control vs optimal control

is lassical control theory still used anywhere in industry? or is it all state-space, digital optimal control/robust control, H-infinity and all that crap?

is there still a point in learning the classical control theory?

Reply to
mr sparkle
Loading thread data ...

It depends on your definition of "lassical control theory"...

Cameron:-)

Reply to
Cameron Dorrough

Yes, nearly everywhere.

Not all, just occasionally here and there.

I guarantee you won't understand any of the other stuff without it!

Kelvin B. Hales Kelvin Hales Associates Limited Consulting Process Control Engineers Web:

formatting link

Reply to
Kelvin Hales

I assume that by "classical" control you mean "classic" methods like lead-lag compensation and PID tuning. The term modern control is usually used to denote methods like LQG, H2 and Hinf.

I understand that the more sophisticated methods, the better model of the plant to be controlled is needed.

There are application areas, where the availability of a good mathematical model is rare. For instance, process control. If you are to design a controller for a mixing valve in a house heating sestem, then standard PID tuning procedures will be a reasonable choice. I cant imagine developing a FEM model for convection heat transfer for a house...

On the other hand, there are application areas, like astronomic telescopes control, nanopositiononing, hard disk drives control, where the physical system can be fairly precisely desribed by a set of differentical equations and then it is reasonable to use this knowledge and fit the controller better to the system by using various sophisticated optimization methods. Improvement on the order of percents might be appreciated when it brings money.

Best regards, Zdenek

Reply to
Zdenek Hurak

{edited}

And, just in case this has been forgotten by the younger generation of Control Engineers: PID control is optimal for 2nd-order systems. The common use of PID control in the process industries is based on the principle that no matter how complex the process, it can usually (no, not always) be sub-divided for control purposes into a hierarchy of smaller systems that have predominantly 1st/2nd-order dynamics. Furthermore, the control usually *is* partioned like this, so that human operators can understand the behaviour and intervene if they wish.

Of course there is a small role for fancier systems using modern design techniques; but in my experience, they tend to turn up in 'black-box' applications, such as in turn-key equipment, where no-one but the designer needs to know what goes on inside the box.

Kelvin B. Hales Kelvin Hales Associates Limited Consulting Process Control Engineers Web:

formatting link

Reply to
Kelvin Hales

-- snip --

Engineers: PID

it can usually

smaller systems that

*is* partioned

if they wish.

techniques; but in my

turn-key equipment,

And they generally work best when the control system designer can influence mechanical and circuit design decisions. In my experience I often add the most value to a control system design by working with the team that's designing the mechanics; the actual control algorithm is so heavily constrained by the nature of the plant that if it ain't right going in there's nothing to be done with control rules.

Reply to
Tim Wescott

Yes, the vast majority or all control loops are 'classical' control. The question is a bit like asking, "Are wheels still used anywhere in industry or is it all spaceships, super critical torpedoes and all that crap? Is there any point at all in learning Newtonian mechanics?"

The basics NEVER go away.

Walter.

Reply to
Walter Driedger

PolyTech Forum website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.