Understanding the Use of Kalman Filters and Gyros in Aircraft Design

[aymank1985]

Hi,
I am starting a project that involve building a aircraft .
I am just not very clear about using kalman filters, gyros and Magnetometer.
From what I read , the kalman filter is used to filter the drift in gyros reading and combine it with the accelerometer reads .
In our project we will be using a gyros and magnetometer, will i still be able to use kalman filter with gyro only ?
Hope someone can clear this to me.

Thanks

 

[Antiphon]

You don't need any of that to build an airplane.

 

[aymank1985]

We are building a on board computer system for a satellite.

 

[Antiphon]

We are building a on board computer system for a satellite.

That sounds like a very interesting project. The good news is you don't need magnetometers in space.

How will your satellite get into orbit?

 

[aymank1985]

How is that ? We are planing to use the magnetometers will be used to detect the position of the satellite using the IGRF modeling .
The satellite won't be lunched in the space , but we will be simulated .

 

[D H]

Have you, or anyone on your project, taken the pretty much standard signals and systems (aka signals and systems, linear systems and signals, or some other variation) class? You need that before you can even start understanding a Kalman filter.

There's nothing wrong with an attitude-only filter. The magnetometer will give a rather coarse and incomplete state update. Without it, you would be doing dead reckoning. The integrated attitude would drift (random walk) with respect to truth over time.

One problem with a magnetometer is that you need to know where you are to know what the reading means. How are you estimating your translational state?

 

[aymank1985]

Yes , I have done system and signals courses .
For the TS we will be using a c++ scripts that will take care of converting orbit elements to position and velocity vectors

 

[D H]

Some questions then. The answers will help me to help you.

1. Environment.
To what fidelity are you simulating the environment? Is gravity spherical or non-spherical? Do you have third body effects? Atmospheric effects? To what fidelity are you modeling the atmosphere, and atmospheric drag?

With regard to rotational influences, are you modeling gravity gradient torque or torque due to atmospheric drag? Magnetic influences on orientation (if any)? Since you have a magnetometer, you need to be simulating the magnetic environment as well. Modeling the Earth's upper atmosphere and magnetic field can vary from very simple to getting real hairy real fast.

2. Sensors and effectors.
To what fidelity are you simulating your sensors and effectors (assuming you have effectors)? Can they be misaligned (in your sim)? Is there some element of randomness to them?

3. State representation.
How are you representing orientation? Direction cosine matrices, Euler angles, quaternions, Rodrigues parameters, modified Rodrigues parameters?

4. Avionics.
What is your sampling rate? Your flight software rate? How accurate are your simulated sensors? What kind of avionics architecture does your simulated spacecraft have (and do you care)?

5. Yourself.
What grade level project is this? How long will you be working on this project? Do you have an advisor who is helping you out?

 

[aymank1985]

Well I think some of those questions I don't have answers to .
Basically we will be building a cage that can generate the same force around the Earth TIMES X ( This will increase the ability to use a magnetometer with a lower resolution and lower cost ) We will use HMC5883L .
I am just not sure if I have the pieces in the right order .
At T=0
1-I should use the SGP4 model with initial TLE , which will supply me with position and velocity vectors .
2-The position vector will be used to compete the IGRF model , which will supply the magnetic field in the ECEF .
3-The IGRF readings need to be converted to ECF then to body frame .
4-The readings from the magnetometer is obtained.
5-The two data are supplied to a EKF .

Please let me know if I have that in the right order , and if there is anything you wan to point out .

Thanks alot

Some questions then. The answers will help me to help you.

1. Environment.
To what fidelity are you simulating the environment? Is gravity spherical or non-spherical? Do you have third body effects? Atmospheric effects? To what fidelity are you modeling the atmosphere, and atmospheric drag?

With regard to rotational influences, are you modeling gravity gradient torque or torque due to atmospheric drag? Magnetic influences on orientation (if any)? Since you have a magnetometer, you need to be simulating the magnetic environment as well. Modeling the Earth's upper atmosphere and magnetic field can vary from very simple to getting real hairy real fast.

2. Sensors and effectors.
To what fidelity are you simulating your sensors and effectors (assuming you have effectors)? Can they be misaligned (in your sim)? Is there some element of randomness to them?

3. State representation.
How are you representing orientation? Direction cosine matrices, Euler angles, quaternions, Rodrigues parameters, modified Rodrigues parameters?

4. Avionics.
What is your sampling rate? Your flight software rate? How accurate are your simulated sensors? What kind of avionics architecture does your simulated spacecraft have (and do you care)?

5. Yourself.
What grade level project is this? How long will you be working on this project? Do you have an advisor who is helping you out?