In this paper, a Jaya-optimized feed forward neural network (FFNN) is used to model the brushless DC (BLDC) motors in the quadcopter. Precise modeling of motor-propeller performance parameters for a given throttle input, including speed, thrust, current, torque, and power is necessary for estimating energy consumption and optimizing energy sources for various flying conditions. Accurate modeling is essential for performance analysis, but it is challenging due to the nonlinear relationships, environmental factors, and variability in propeller and motor characteristics. However, by integrating experimental data with simulation and analytical models, the prediction accuracy can be enhanced. While FFNNs are effective in capturing the complex nonlinear relationships between input and output variables, their performance is highly dependent on the optimization of network weights and biases. Traditional optimization methods often face challenges such as overfitting, computational inefficiency, and sensitivity to initial conditions. By leveraging the Jaya algorithm, which requires minimal control parameters and promotes adaptive convergence, the FFNN’s predictive accuracy is significantly improved. In this research, the performance of the model is studied and verified experimentally with two different throttle conditions, namely 1). linearly increasing and 2) full flight from takeoff to landing. The following performance parameters speed, thrust, current, torque, and power are predicted with an accuracy of 96.4%, 97.5%, 97.2%, 94.9%, and 96.8%, respectively, at maximum throttle positions with respect to traditional models.
