Regularized Regressions-Based Data-Driven Modeling of BLDC Motor Characteristics for UAVs

Precise estimation of thrust and torque is essential for the efficient design and control of brushless DC (BLDC) motors, particularly in applications such as unmanned aerial vehicles (UAVs). The inherent nonlinear dependencies between system inputs and corresponding outputs present significant modeling challenges. In this work, a data-driven approach is adopted using regularization-based regression methods, namely LASSO, ridge, and elastic net, to predict BLDC motor thrust and torque. The investigation was carried out on a standardized test bench developed at IITRAM, Ahmedabad, where the acquired experimental data were divided into distinct training and testing subsets. The models were trained on the former and assessed on the latter to evaluate predictive performance. Among the techniques studied, the elastic net approach demonstrated the highest predictive accuracy under varying operational conditions, effectively balancing bias and variance. These findings underscore the potential of regularized regression frameworks as reliable tools for modeling BLDC motor behavior in UAVs and related electromechanical applications.