2DOF-Helicopter-NARX_BPTT_Adam

Overview

This project models the nonlinear dynamics of a 2-DOF helicopter system using a Nonlinear AutoRegressive model with eXogenous inputs (NARX) neural network. The model predicts the system's behavior based on lagged input-output features, leveraging temporal dependencies to output pitch and yaw angles.

Colab Link

You can access the Google Colab notebook for this project:

Data Used

The model is trained and validated using real sequential data representing the helicopter's behavior:

• Input Variables:
  • V: Voltage applied to the motors (constant for both motors).
  • I_pitch: Current for the pitch motor.
  • I_yaw: Current for the yaw motor.
• Output Variables:
  • pitch_angle: The pitch angle of the helicopter.
  • yaw_angle: The yaw angle of the helicopter.

Data Format

The input and output data are stored in separate CSV files:

• inputs.csv: Contains columns V, I_pitch, I_yaw.
• outputs.csv: Contains columns pitch_angle, yaw_angle.