PID control of inverted pendulum using PLC

hi all!
m doing my FYP related to PID control of Bytronics inverted pendulum.First of all m little bit confused that whether it is possible or not!
I did analog scaling in Allenbradeley successfully.I know how 2 use PID also!.but the thing which i wan know is that whether it is possible to control non-linear system like inverted pendulum where more than one gains are to adjusted.
The sensors with which i m dealing with are:
1-Linear position Potentimeter
2-Angular position potentiometer
3-Tachometer
m attaching manual with complete information about inverted pendulum.plz help me about this:

zahidraza1212 said:

hi all!
m doing my FYP related to PID control of Bytronics inverted pendulum.First of all m little bit confused that whether it is possible or not!

Click to expand...

I have seen PLCs control inverted pendulums at trade shows as far back as the 1990s.

someone last year tried to use an AB SLC-500 to control a single inverted pendulum but found the reaction time wasnt fast enough and ended up using Labview with a DAQ card instead.
I suppose it will depend heavily on the CPU u are using!!

LH

Originally posted by Peter Nachtwey:

I have seen PLCs control inverted pendulums at trade shows as far back as the 1990s.

Click to expand...

I've seen a Ford Taurus do 200 MPH at Talladega. That doesn't mean I can go down to my local Ford dealership and buy one that will do that on the street. I don't think there are too many plcs on the market today that have the I/O rate to pull this off. Back in the '90's that must have been a pretty select group.

I think alot depends on the initial conditions. Keeping an inverted pendulum inverted is a much easier proposition than making a non-inverted pendulum inverted in the first place.

Keith

Did you guys search Youtube first?

No, but I did
http://www.youtube.com/watch?v=PpLVlFB5fvI&feature=related

Controlling an inverted pendulum is well documented and there are plenty of examples on the internet. It can be done with a little micro controller ( A jab at a previous thread ).

The trick is not programming the controller but designing the inverted pendulum system so that it can be controlled. If the pendulum is heavy or the motor is slow then controlling an inverted pendulum is going to be impossible no matter what controller you are using. Look at the video. The rod is long, the mass is small, and the motor is very fast.

I bet the person who failed with the SLC 500 had a marginal system or didn't really know how to control the inverted pendulum.

Do a search for inverted pendulum labview. You will see that the inverted pendulum has been many times with lab view. I would be surprised if National Instruments did have the code posted somewhere on their website as a demo program. That would make implementing an inverted pendulum easy.

I am pretty sure I could do it with my S7 315 by my desk. The scan times are fast enough if that is all you are worried about.

Why does everyone think they are the first to do something? Search the web. I bet 99% has been done before.

I'm personally more fond of this one:

http://www.tecnun.es/asignaturas/control1/proyectos/pdobleinv/imagpdoble/PenduloDoble.mpg

I came across it a couple of years ago and thought it was pretty neat. Granted, it is the next step up in complexity since it is a double pendulum. But you can see what I mean about the difference between maintaining the inverted state and getting to he inverted state in the first place.

Originally posted by Peter Nachtwey:

It can be done with a little micro controller

Click to expand...

I would believe that. But a microcontroller isn't burdened by all the added 'stuff' that makes a plc the thing maintenance techs the world over know and love. The more dedicated you can make a controller the better the performance for the selected task (all other things being equal).

Apparently my analogy was lost. It wasn't about power. It was about availability. Just because someone shows you something at a show doesn't mean you can go to the corner store tomorrow and buy it. The Detroit Auto Show is one big example of 'you will never get this anywhere'.

Originally posted by Peter Nachtwey:

I am pretty sure I could do it with my S7 315 by my desk. The scan times are fast enough if that is all you are worried about.

Click to expand...

This discussion has come up in several varieties before. It is seldom simply about scan time. I can pretty easily get sub-millisecond scan times out of a current generation CLX processor. Does that mean I should be able to do motion control with that processor without a motion card? Hardly. It is often more about getting the information into and out of the platform than it is about raw processing power. Just as important that is scan jitter. What good does my one millisecond scan time do me if my I/O can't respond in less than 30 milliseconds and has two milliseconds of jitter on top of that? Yes, that's rhetorical.

As I said, if it starts upright you have a shot if you look at your I/O carefully enough. If you have to swing it up I don't think a plc with standard I/O will be fast enough.

Keith

I doubt that double pendulum was done with a PLC.

Keith has also made a point as to why so many motion control projects fail when trying to use a PLC only implementation. Still, the PLCs manufacturers have fast analog I/O if you are will to pay for it but it is a joke. AtoD conversions should take no more than a few micro seconds. Even back in the mid 90s the AtoD conversion were that fast and same goes for the DtoA.

However, if one is clever they can take into account the delays of the AtoD and DtoA but this isn't the kind of thing most PLC people can do. However, the best and brightest of a PLC company should be able to do it. I don't know where that leaves the OP.

Originally posted by Peter Nachtwey:

However, the best and brightest of a PLC company should be able to do it. I don't know where that leaves the OP.

Click to expand...

Very careful selection of hardware.

Keith

m very thankful to all