Where to start? Modeling a UAV Control System in Simulink

[Tomo8281]

Hi,
I have a small project coming up, based around modeling (and perhaps building) a small Micro UAV system. Its a small device with contra-rotating blades (for stability), and I want to perhaps model the control system using simulink but I really don't know where to start! The aim is for the system to be able to hover and stabilise itself. I've used Matlab a lot in the last 4 years, and simulink a couple of times, but I need a bit of pointing out what to start with and how to go about doing this.

Can anyone point me in the right direction? I've spent the past 3 weeks trying to find a starting point but I'm a bit stumped.

Thanks

 

[Floid]

I would start with modeling the plant. In your example I would probably:

Implement the equations for the lift of your blades from their velocity. Test it to make sure it works.

Then progress into adding the equation for the velocity of the blades given an electrical input (model your electric motor and gears). Test this part and then integrate it with the lift equation.

Depending on how complex your model is just these parts can get to be tricky. Especially if you start taking into account the moment of inertia of the blades, wind resistance, non-linearities in the torque of the electric motor. But start simple and then build in the complexity. So start assuming the blade always counter rotate in sync, that everything acts linear, etc. Then once you are satisfied that works start incorporating more and more complexity to whatever point is deemed appropriate.

Now model whatever feedback devices you will be using for your UAV model. Do you control based on some type of altitude reading, tachometer, etc?

The final step would be putting in whatever your control algorithm is going to be. Do this after you have individually tested all the other components of the model to make sure they work.

 

[Tomo8281]

I would start with modeling the plant. In your example I would probably:

Implement the equations for the lift of your blades from their velocity. Test it to make sure it works.

Then progress into adding the equation for the velocity of the blades given an electrical input (model your electric motor and gears). Test this part and then integrate it with the lift equation.

Depending on how complex your model is just these parts can get to be tricky. Especially if you start taking into account the moment of inertia of the blades, wind resistance, non-linearities in the torque of the electric motor. But start simple and then build in the complexity. So start assuming the blade always counter rotate in sync, that everything acts linear, etc. Then once you are satisfied that works start incorporating more and more complexity to whatever point is deemed appropriate.

Now model whatever feedback devices you will be using for your UAV model. Do you control based on some type of altitude reading, tachometer, etc?

The final step would be putting in whatever your control algorithm is going to be. Do this after you have individually tested all the other components of the model to make sure they work.

I do not have a physical model of the device, I only have a simple schematic (dimensions). The propellers are mounted on a sort of gimballed platform that can rotate, which will be used to provide an offset off the thrust axis for our movement control. Apart from that, it is a simple Micro coaxial helicopter design.

So would I need to actually design and build the system first, before I can do any simulation model?

 

[Floid]

Well, this is where modeling a system, especially an aerodynamic system, becomes a very intensive effort.

Start with just the hovering case so you can ignore the tilt of the rotor and assume all lift generated by it is in the vertical direction.

You should be able to find/measure the dimensions of the rotor blades, the pitch of the blades, the mass of the aircraft, etc. From this you can figure out what amount of lift necessary for the helicopter to hover and then the necessary rotation rate for the rotor to generate this lift.

Then you need to research/test/make assumptions about your motor/gears. The easiest may be to use a voltage source to apply known voltages to it and measure the rotation rate of your rotor.

You just have to slowly work and build up your model...

 

[viscousflow]

So would I need to actually design and build the system first, before I can do any simulation model?

No, this is why simulators were invented.

I find it amazing no one has suggested starting with the basic equations of motion first. This is actually typical and should be the only starting point. After you get those down, you then look at each term in detail (As per the above suggestions). Google uav thesis papers, these are good guides. This way you will have an intuitive grasp of your system.

 

[Tomo8281]

Ok thanks for the help so far, I think I'm getting it. I'm currently working on the rotational transform matrix, around that area.

One of my main problems is though that the MAV system has a concept of using gimballing of the rotor head to provide offset of the thrust axis from the centre of gravity to provide restoring forces to effect stability and lateral flight control. I'm trying to put together a system model but I am having a hard time since I don't know how to account for the tilting of the rotor head for flight control? I can't find any research papers of other people using the same idea :/!

 

[viscousflow]

One of my main problems is though that the MAV system has a concept of using gimballing of the rotor head to provide offset of the thrust axis from the centre of gravity to provide restoring forces to effect stability and lateral flight control. I'm trying to put together a system model but I am having a hard time since I don't know how to account for the tilting of the rotor head for flight control?

So basically you're allowing the head to wobble to maintain roll stability? Why would you do that?

Accounting for that is something called gyroscopic motion.

 

[Tomo8281]

So basically you're allowing the head to wobble to maintain roll stability? Why would you do that?

Accounting for that is something called gyroscopic motion.

It is a new idea I've been tasked to design. Hence why I'm asking for help :)! I'm realizing I may not be able to obtain stabilization from it, only a control mean... Does this sound right?

 

[Tomo8281]

Still working on this. I'm finding it hard to prove that this will not work (so I can go on and design a better system). I don't know how to 'add' the tilting center of gravity to any control equation or the AB matrix.

If I can find out how to add this to a numerical control matrix (for example the AB matrix) I believe I can use MATLAB to show that it will not work.

The reasoning behind the gimballed device was that from a consideration that if the platform was given pendular stability (centre of gravity placed well below the thrust generating rotor head) the thrust axis could be tilted with respect to the centre of gravity to provide stabilisation and a side force with respect to upsets caused by side winds and gusts.

 

[viscousflow]

Add a pendulum to your model with a similar weight and inertia as the original counterweight. You will have to re-derive the equations of motion.