I'm trying to create an algorithm (neural network) that is able to predict a time series from a set of different parameters that are not given through time. Let's say I have a plane flying under the following conditions:

Parameters Value
Angle of attack 8 degrees
Lateral angle 12 degrees
Wind speed -20 m/s
Plane speed 200 m/s

From this point, I would like to predict the translational velocities in x-y-z axis for the next 2-3 seconds.

In order to train my model, I have a data base with different initial situations (input) and different motions of the plane (desired output) linked to their initial situation. Therefore, I want to train my model to predict these motions mentioned before, based only on the initial situation described.

In other words, the basics of what I'm trying to do could be summed up as the following:

Parameters describing the initial situation -> Model -> Time series of translational velocities.


I don't think you need a recurrent neural network for this. Why not just train a feedforward model with angle of attack etc as input and translation velocities as output? The size of the output will depend on how frequently you want updates to the velocities. e.g. if the update rate is 0.5 seconds and the network is predicting 2 seconds in advance it could have 12 outputs, being x,y,z values every 0.5 seconds.

You may need to massage the way you represent the outputs to find the best approach. For example, will the network be predicting absolute velocities or just the initial velocities plus changes to velocity over time? If the latter you can always get back to absolute velocities.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.