Experimental and data-driven modeling based analysis of upstream wall-proximity effects on the performance of propellers for multirotor UAVs

Abstract

The present study investigates the influence of upstream wall proximity on thrust and efficiency in multirotor UAV systems through a combination of experimental measurements and data-driven artificial neural network (ANN) based modeling approaches. Experiments are conducted using a test stand system with propeller diameters ranging from 9 to 13 inches, for motor speeds of 1500–6000 RPM and normalized upstream wall distances (h/R = 0.25–2). The results show that upstream wall proximity leads to measurable changes in thrust and efficiency. Thrust, which governs lift generation for multirotor UAVs and payload capability, exhibited significant variation at low h/R values. In contrast, efficiency, which determines energy utilization and flight endurance, showed comparatively smaller variations, indicating a weaker dependence on upstream wall proximity. To complement the experiments, an ANN model trained using the scaled conjugate gradient (SCG) algorithm is developed for performance prediction. The model achieved high predictive accuracy, with coefficients of determination (R 2 ) of 0.99 for thrust and 0.98 for efficiency. The study provides a validated experimental database and a reliable data-driven modeling framework and the findings contribute to improved understanding of confined-flight aerodynamics and support the development of efficient UAV design and control strategies for indoor and infrastructure-based applications.