Interesting New Controller Doesn't Need Tuning???

Watch video.

http://www.youtube.com/watch?v=N_zonudtKEM&feature=colike

My Dad always said if something sounds to good to be true it normally is. I would not mind testing one of these. What is the catch? Maybe cost a fortune.

I would also be interested to know the algorithm behind this. Anyone got some thoughts?

I sent an Email requesting more info..be interesting to see what the send.

I made some comments on Youtube video. It is easy to see what they are doing. Notice that after the first cycle the control output seems to change about half of what is needed and then waits for the PV to respond and if the response is close to what is predicted then the the control output goes to the estimated value for the SP. All they are doing is estimating a feed forward by using a 'guessing' algorithm. It is kind of like trying to guess an number between 0 and 100.

I discout videos like this where the PID is allowed to look bad. It tells me they DON'T know how to tune a PID and certainly don't know control theory. I have seen lots of people claiming to have a better control method but their better control method only looks good if the PID looks bad in comparison.

RheinhardtP said:

My Dad always said if something sounds to good to be true it normally is. I would not mind testing one of these. What is the catch? Maybe cost a fortune.

I would also be interested to know the algorithm behind this. Anyone got some thoughts?

Click to expand...

Probably not the same, but here is a pdf on root recursion and process optimalization without parameters. It is in German, but the equations are in Greek (to me, at least ).
http://www.me.th-mittelhessen.de/fileadmin/download/orlowski/Prozessoptimierung-ohne-Parameter.pdf

Kalle

I don't know PID control.
I like to see stuff like this from a sales l pitch view. If I am trying to sell you a truck and I show you 1 new clean polished and shiney or I show you the exact same make and model just covered in dents, rust and dirt. What looks better?

My guess is that none of the algorithms were tuned when the process times changed. So to me it looked very promising.:nodi:

Kalle

"Self Tuning" PID almost always involves some perturbation of the process. The tuning constants are usually calculated using Ziegler Nichols or some other semi-empirical method after either: changing the loop’s output and then measuring the process variable’s response, or changing the setpoint and watching the process response. I think an experienced engineer like Peter can probably do a better tuning job than the self tuning controllers.

I am constantly amazed that so many engineers use PID control without considering alternatives. A fairly large industry has developed aroung PID tuning books, tuning software, tuning magazine articles, self-tuning controllers, fuzzy logic modifications to PID, and so on and on and on. It strikes me that an algorithm that is that tough to tune isn’t really a successful approach! Coupled with the studies that estimate at least half the PID loops installed are actually being operated in manual, it makes you wonder.

I suspect that most engineers just use it because it is in the books and they were taught it in school, mostly by professors who have never tuned a loop in a real industrial setting. Besides, the equations look really cool!

JeffKiper said:

I don't know PID control.
I like to see stuff like this from a sales l pitch view. If I am trying to sell you a truck and I show you 1 new clean polished and shiney or I show you the exact same make and model just covered in dents, rust and dirt. What looks better?

Click to expand...

Neat n shiny - yup looks good - except there is no engine ??

I will take the truck dents and all - mostly because it has an engine in it and the engine works. Also would be willing to take a discount for dirt n dings. I need a truck to haul not look pretty.

IF their product can hook up and run the load properly with no tuning that would be great.

Seems to me if you are relying on PID you should just accept the fact you are going to have to tune and get things "tweaked in". The really horrible part to advocates of Plug n Play is they have a learning curve in front of them.

I would not trust the video and would expect them to come out and show me in person.

Dan Bentler

Uh, yeah, I am thinking they have a little way to go yet . . .
http://acorient.com/

German, but the equations are in Greek (to me, at least ).

Click to expand...

Math is like that. It is its own language.

I like to see stuff like this from a sales l pitch view.

Click to expand...

But you will be fooled.

I think an experienced engineer like Peter can probably do a better tuning job than the self tuning controllers.

Click to expand...

This depends on the algorithm they use. I write auto tuning algorithms and I know there are many techniques and refinements to the techniques that aren't in the books. ZN is rarely used by any auto tuning algorithm. One can do so much better than ZN without much effort.

I am constantly amazed that so many engineers use PID control without considering alternatives.

Click to expand...

I think it is mostly a documentation issue but on top of that the basic principles of PID control is that you can place the poles to get the response you want. It takes a little math though. I have little respect for those methods that don't attempt to model the system in some way. ZN, Fuzzy logic and some others are like this.

I like Sliding Mode Control. However, it doesn't do well when there is a long dead time UNLESS one adds a Smith Predictor. Then the performance is very robust.

I suspect that most engineers just use it because it is in the books and they were taught it in school, mostly by professors who have never tuned a loop in a real industrial setting.

Click to expand...

I can point to examples in books where the professor goes overboard on the theory. It is obvious he hasn't used the technique in a real application.

I would not trust the video and would expect them to come out and show me in person.

Click to expand...

I think the video is valid but it is setup for just one system. The system is non linear but other than that it is pretty simple. What if there is a dead time, what if there are underdamped poles? These issues weren't address. If you noticed I complained about their term 'delay'. What delay? They don't make that clear.

Link to a patent reveals some information:

http://www.patentbuddy.com/Patent/6781096

OkiePC said:

Uh, yeah, I am thinking they have a little way to go yet . . .
http://acorient.com/

Click to expand...

That kinda tells me all in need to know about this product.

Tom Jenkins said:

"Self Tuning" PID almost always involves some perturbation of the process. The tuning constants are usually calculated using Ziegler Nichols or some other semi-empirical method after either: changing the loop’s output and then measuring the process variable’s response, or changing the setpoint and watching the process response. I think an experienced engineer like Peter can probably do a better tuning job than the self tuning controllers.

I am constantly amazed that so many engineers use PID control without considering alternatives. A fairly large industry has developed aroung PID tuning books, tuning software, tuning magazine articles, self-tuning controllers, fuzzy logic modifications to PID, and so on and on and on. It strikes me that an algorithm that is that tough to tune isn’t really a successful approach! Coupled with the studies that estimate at least half the PID loops installed are actually being operated in manual, it makes you wonder.

I suspect that most engineers just use it because it is in the books and they were taught it in school, mostly by professors who have never tuned a loop in a real industrial setting. Besides, the equations look really cool!

Click to expand...

I agree with you Tom. I think this is one of the most misused functions in our industry. I hardly ever see properly tuned PID loops and like you said half of them get run in manual.

Rheinhardt

Peter,

Saw your comments on YouTube...can connect you to Acorient directly for answers to your questions. Try [email protected]

My understanding is Acorient has been 'under the radar' for a couple of years as they finalize development of their architecture (non-algorithmic) and platform. PID replacement is first of several applications in plan.

btw: PID was tuned by expert third party for the vid.