Power consumption model for Unmanned Aerial Vehicles using Recurrent Neural Network techniques

Unmanned Aerial Vehicles (UAVs) have become increasingly integral across diverse sectors, necessitating accurate power consumption modeling to optimize flight operations and ensure reliability. Traditional approaches often fail to capture the intricate, non-linear dynamics between operational parameters and power usage. This study introduces deep learning techniques, including RNN, GRU, LSTM, Bi-LSTM, and SA-Bi-LSTM, with various activation functions and optimizers for predicting UAV power consumption using an extensive dataset. Additionally, the influence of activation functions and optimization algorithms on model performance is assessed. Bi-LSTM demonstrates superior predictive accuracy, as evidenced by RMSE and MAE metrics.