Track machine auto-leveling system

I am working in a challenging project to program an auto-leveling system for a big track machine.
The system has to be designed to limit the tilt while tramming by adjusting support height in response to varied terrain. Automatically move to near center-stroke before tramming.

Controls
- 4 hydraulic cylinder, one for each side of the machine. The two cylinder on the front are hydraulically connected together then will be a total of 3 hydraulic directional valves instead of 4.
- 1 hydraulic directional valve for each cylinder, 2 electrical coils per valve (up and down) = total 6x coils - 6x DO
- Tilt-sensing: it will indicates pitch (4-20ma) and roll (4-20ma)
- The hydraulic valves are 3 positions center closed 4 way.

I am not sure if a 3 step PID will apply or not. Since there will be 2x[input_per] pitch and roll what will be my feedback?
How I am going to avoid the PID [pitch] and [roll] fight each other? . . . or maybe just some comparator operations?

Does anyone have any idea?
Thanks in advance.

you think wrong, you only have two directions and 4 cylinders.

just have one cylinder for example the rearleft in manual halfway position.
the rearright will be controlled by roll and the front by the pitch.
a 3 step controller without PID will work, so if pitch pos then lower front wheels etc.
just some comparators. as movement is not controlled in anyway.

better is to use proportional valves.

shooter said:

you think wrong, you only have two directions and 4 cylinders.

just have one cylinder for example the rearleft in manual halfway position.
the rearright will be controlled by roll and the front by the pitch.
a 3 step controller without PID will work, so if pitch pos then lower front wheels etc.
just some comparators. as movement is not controlled in anyway.

better is to use proportional valves.

Click to expand...

Thanks Shooter.
Great idea, i like it!
I will have to leave with directional control valves since the hydraulic system was designed by someone else.
Let me see if I understood correctly your idea: with regards the rear left in manual, I can not do it since the requirements is full automatic-leveling system....while machine moving the system has to maintain its gravity center at 1º even. But I like the idea of using pitch signal to control front and roll to control side cylinders. Still trick because one will be affected by the other.
What do you mean 3 step controller without PID?
Thanks for helping me out.

I agree it would be much easier with proportional valves. Do you at least have feedback from the cylinders? I am a bit confused by the "two cylinder on the front are hydraulically connected together" part. Do the two cylinders share a common pivot at the top? I'm picturing a triangular layout....

If I understand correctly, then roll could be controlled by moving the rear cylinders in opposite directions simultaneously (i.e. left UP and right DOWN to roll right). If you only move one, you limit how far you can roll. When you run out of travel, you would have to move the other one. Therefore, I see a simultaneous movement being simpler in the long run.

Similarly, pitch could be controlled by moving the front cylinder (singular, since they're coupled) plus BOTH rear cylinders in opposite directions simultaneously (i.e. front UP and BOTH rear DOWN to pitch up).



-Eric

Eric Nelson said:

I agree it would be much easier with proportional valves. Do you at least have feedback from the cylinders? I am a bit confused by the "two cylinder on the front are hydraulically connected together" part. Do the two cylinders share a common pivot at the top? I'm picturing a triangular layout....

If I understand correctly, then roll could be controlled by moving the rear cylinders in opposite directions simultaneously (i.e. left UP and right DOWN to roll right). If you only move one, you limit how far you can roll. When you run out of travel, you would have to move the other one. Therefore, I see a simultaneous movement being simpler in the long run.

Similarly, pitch could be controlled by moving the front cylinder (singular, since they're coupled) plus BOTH rear cylinders in opposite directions simultaneously (i.e. front UP and BOTH rear DOWN to pitch up).



-Eric

Click to expand...

Thanks Eric.
Yes Sir, I agree with the proportional valves as well, but it should work with 4 way, 3 positions, center close directional valves.
All cylinders has linear transducer providing 4-20mA.
With regards the front cylinders, each one has its own place, they are not connected mechanically, right front cylinder and left front cylinder...however both will be controlled by the same valve....hope this helps. That is why I mentioned about 3 directional valves as total: front cylinders valve, right rear valve and left rear valve.
Since there are linear transducer in the cylinders i was thinking if I could use them as my feedback? then the 3 step PID may works.
Thanks.

This is a disaster waiting to happen

First, the control method should be worked out before deciding how to design the mechanics and hydraulics.

Is there a fixed pivot point somewhere in the middle so the upper platform will roll and pitch around this fixed point. That hasn't been mentioned. Four cylinders by themselves are not stable unless the cylinders themselves are fixed to the lower platform. If the cylinders can't pivot there will be a bending action of the rods as soon as you start to tip the upper platform relative to the bottom.

I would approach this problem like it is a flight simulator with limited, two, degrees of freedoms. A flight simulator usually has 6 degrees of freedom. I am not sure I like the idea where the front cylinders are controlled by only one valve.

I would use servo quality proportional valves. I would have a PID for each servo valve. There needs to be a master supervisor program that reads the pitch and roll feedback and computes the end points for the cylinders.

I understand why the mechanical guys want to control the two front cylinders with one valve but it makes controlling the two difficult because they can't be controlled independently.

Still trick because one will be affected by the other.

Click to expand...

Yes, and there will be a lot of trig to work this out. Then you will need another algorithm that will move your from where the actuators are to where they need to be.

Have the mechanical and hydraulic designers done this before? Probably not, if so they would have a control program for this already.

BTE, I have a canned 6DOF program worked out. It would need to be simplified to 4 actuators to work with your system and I would use two servo valves for the front two cylinders.

I know others here are way beyond me in the control algorithms. I'm just confused. If both sides of the front are controlled by one valve how is the front adjusted in a roll situation?

The only thing I can envision is this:

Lower track component

Coupling to an intermediate section pivoting on a roll axis support, this part moved by two cylinders.

Coupling to an upper section pivoting on the pitch axis and controlled by the front cylinders.

Maybe I need a diagram.

bernie_carlton said:

I know others here are way beyond me in the control algorithms. I'm just confused. If both sides of the front are controlled by one valve how is the front adjusted in a roll situation?

Click to expand...

The back two cylinders will control the roll. The front pitch needs to be computed by the average of the two cylinders. The question is why have the two cylinders up front and simply make a tripod? Four cylinders can bind if the calculations are done right but controlling the two front cylinders with only one valve means that the back cylinders can twist the platform and the oil in the front will flow where it needs to because the pressures on both sides are equal. The problem is that this scheme makes it much harder for the controller.

The only thing I can envision is this:

Lower track component

Coupling to an intermediate section pivoting on a roll axis support, this part moved by two cylinders.

Coupling to an upper section pivoting on the pitch axis and controlled by the front cylinders.

Maybe I need a diagram.

Click to expand...

Too complicated. I would use one bigger cylinder up front and keep it simple since two front cylinders without individual valves will do nothing for the roll control anyway. Having one bigger cylinder in front simplifies the control.

I still want to know what keeps the upper platform stable. If the is a pivot point near the center of the upper platform then then calculations for the actuator position will have to be precise to keep the platform from bending around the pivot point. How is one going to do that in a PLC? One would have to rotate vectors in 3D from the pivot point.

Peter Nachtwey said:

First, the control method should be worked out before deciding how to design the mechanics and hydraulics.

Is there a fixed pivot point somewhere in the middle so the upper platform will roll and pitch around this fixed point. That hasn't been mentioned. Four cylinders by themselves are not stable unless the cylinders themselves are fixed to the lower platform. If the cylinders can't pivot there will be a bending action of the rods as soon as you start to tip the upper platform relative to the bottom.

I would approach this problem like it is a flight simulator with limited, two, degrees of freedoms. A flight simulator usually has 6 degrees of freedom. I am not sure I like the idea where the front cylinders are controlled by only one valve.

I would use servo quality proportional valves. I would have a PID for each servo valve. There needs to be a master supervisor program that reads the pitch and roll feedback and computes the end points for the cylinders.

I understand why the mechanical guys want to control the two front cylinders with one valve but it makes controlling the two difficult because they can't be controlled independently.


Yes, and there will be a lot of trig to work this out. Then you will need another algorithm that will move your from where the actuators are to where they need to be.

Have the mechanical and hydraulic designers done this before? Probably not, if so they would have a control program for this already.

BTE, I have a canned 6DOF program worked out. It would need to be simplified to 4 actuators to work with your system and I would use two servo valves for the front two cylinders.

Click to expand...

Thanks Peter,
Yes, the cylinders are fixed to the upper structure on two independent tracks, left and right.
It seems that I am missing something because I don't see now how I am going to adjust the roll if the front cylinders are connected together....I am confused now. I need to double check with mechanic engineers.
This is a full new design...we actually never designed anything huge like this before.

What is canned 6DOF? Is that a program that you can send to me to check?
Regards.

I have to ask the question
- I assume you want to level a field or similar.
it is ok for you to work this out and program it.
BUT
as laser levelling equipment for hydraulic machinery is 'Off the Shelf'
why would you?

ftsouza said:

Thanks Peter,
Yes, the cylinders are fixed to the upper structure on two independent tracks, left and right.

Click to expand...

This makes things even more complicated because the distance between the pivot point and the where the ends of the cylinders are in the tracks will change as a function of the tilt angles.

It seems that I am missing something because I don't see now how I am going to adjust the roll if the front cylinders are connected together....I am confused now. I need to double check with mechanic engineers.

Click to expand...

There is only one valve so the oil will flow between the two cylinders. Like I said above, the front cylinders have no control over the roll only the average position needed for pitch control. This part will work mechanically and hydraulically but it will make your job MUCH harder. I have seen cases where the poor control guy gets stuck with problems that are too difficult for him to solve and it ends up wasting weeks or months which in the end cost much more than what it would cost to do the hydraulics and mechanics so the system can be controlled easily.

This is a full new design...we actually never designed anything huge like this before.

Click to expand...

I can tell. On Monday you should ask for all the equations necessary for computing the position of each of the cylinders as a function of pitch and roll and see if you get more than a blank stare other some comment like that that is your job. I have been involved with too many jobs like this where the mechanical and hydraulic 'designers' kludge together a system without the thought of how it is going to be controlled.

What is canned 6DOF? Is that a program that you can send to me to check?
Regards.

Click to expand...

We can control platforms like this:
http://www.youtube.com/watch?v=15ZB0c9ev98
This platform was used in the filming of the movie "2012".
This is controlled by one of our motion controllers.
These 6DOF platforms are used for flight and race car simulators. Some are used for serious training and some just for amusement rides.
What I mean by "canned" is that the program that figures out how to move the 6 axes to a yaw, pitch and roll, x, y and z is done so a lot of the math is done and executed in the motion controller. However, there still needs to be a program that figures out what the roll and pitch need to be in your case but that too can be done in the motion controller.

Another huge system
http://www.entertainmentengineering.com/v2.issue04/index.html?page.15.html
We made the motion controllers for this system. The original controllers and people doing the control failed to control the platform well enough.

This really isn't the job for a PLC. There is too much math.

Peter Nachtwey said:

This makes things even more complicated because the distance between the pivot point and the where the ends of the cylinders are in the tracks will change as a function of the tilt angles.


There is only one valve so the oil will flow between the two cylinders. Like I said above, the front cylinders have no control over the roll only the average position needed for pitch control. This part will work mechanically and hydraulically but it will make your job MUCH harder. I have seen cases where the poor control guy gets stuck with problems that are too difficult for him to solve and it ends up wasting weeks or months which in the end cost much more than what it would cost to do the hydraulics and mechanics so the system can be controlled easily.


I can tell. On Monday you should ask for all the equations necessary for computing the position of each of the cylinders as a function of pitch and roll and see if you get more than a blank stare other some comment like that that is your job. I have been involved with too many jobs like this where the mechanical and hydraulic 'designers' kludge together a system without the thought of how it is going to be controlled.


We can control platforms like this:
http://www.youtube.com/watch?v=15ZB0c9ev98
This platform was used in the filming of the movie "2012".
This is controlled by one of our motion controllers.
These 6DOF platforms are used for flight and race car simulators. Some are used for serious training and some just for amusement rides.
What I mean by "canned" is that the program that figures out how to move the 6 axes to a yaw, pitch and roll, x, y and z is done so a lot of the math is done and executed in the motion controller. However, there still needs to be a program that figures out what the roll and pitch need to be in your case but that too can be done in the motion controller.

Another huge system
http://www.entertainmentengineering.com/v2.issue04/index.html?page.15.html
We made the motion controllers for this system. The original controllers and people doing the control failed to control the platform well enough.

This really isn't the job for a PLC. There is too much math.

Click to expand...

Thanks again Peter,
Usually, electrical engineers get on the boat running already....they just tell you both mechanic and hydraulic system is done and you have to control it how it is.
I will discuss with the team about it on monday.
Thanks a lot for helping me out.
Regards.

iant said:

I have to ask the question
- I assume you want to level a field or similar.
it is ok for you to work this out and program it.
BUT
as laser levelling equipment for hydraulic machinery is 'Off the Shelf'
why would you?

Click to expand...

Thanks Ian,

Can you please give me some links?

Regards.

http://www.level-best.com/
this is a basic one
i have only done a google search for
hydraulic laser levelling systems
I am sure there are systems that will suit your needs.
My cousin runs large (in my mind - I'm a city bloke) laser levelling systems doing contract levelling of large farms

http://www.agl-lasers.com/products/category/C9

theses devices are application based - but they are purely controlling the hydraulice so not too hard to achieve