Numerical and experimental measures of the Unmanned Aerial system UAS-S4

Measurements of the inertial properties are needed during the design of aircrafts. Furthermore, these knowledge represent one of the most problems to be solved while studying aircraft rotational motion or even designing... [ view full abstract ]

Measurements of the inertial properties are needed during the design of aircrafts. Furthermore, these knowledge represent one of the most problems to be solved while studying aircraft rotational motion or even designing aircraft flight control systems. This is the main reason why accurate methods for computing aircraft inertial properties have received sustained interest over the years.

This paper firstly presents a structural analysis of a drone - the UAS-S4 ETHECATL. Mass, center of gravity position and mass moment of inertia are numerically determined through Raymer and DATCOM statistical-empirical methods, coupled with mechanical calculations. Then, experimental tests are performed using the pendulum method, in order to validate the numerical predictions.

When experimentally determining the mass moment of inertia, the bifilar torsion pendulum is used for the moment vertical axis and the simple pendulum for the moment longitudinal and transversal axes determination. A nonlinear dynamic model is developed for rotational motions about the center of gravity of the system under tests. Equations of motion of this model are then obtained, including the effects of large-angle oscillations, aerodynamic drag, viscous damping and additional mass of air. Then, MATLAB genetic algorithms are used to obtain the moment of inertia values that would validate experimental data with the simulated data in a least square error sense.

Generally in such work, the common technique is to ignore the damping and to linearize the equations of motion to model the bifilar or simple pendulum as a harmonic undamped oscillator. Moreover, the choice of physical parameters of the pendulums is not often based on any rigorously derived criteria, although some reasoning is used to set them up. This work builds on common works by incorporating a higher fidelity of dynamic models of bifilar and simple pendulums and by presenting a method for determining the pendulum physical parameters that would “minimize measurements error variance”.

The experimental data is gathered using three different sensors. One triple axes accelerometer with a full scale range of ±2g and 16384 LSB/g of sensitivity enables us to obtain both pitch and roll angles all during the entire motions when applicable. A triple axes magnetometer with a full scale range of 100µT and a sensitivity of 0.1µT/LSB is used to determine the yaw angle. Because both of these two sensors are a little bit noisy, a three axes gyroscope is used to filter the signal, since the accuracy of the results depends on it. A method to correctly calibrate the three sensors is also discussed.

Finally, this paper shows the experimental results for an object of uniform density, for which the moment of inertia is numerically computed straightforward from geometrical data, and also shows the experimental results obtained for the UAS-S4 ETHECATL, which is the main system on which the new moment of inertia measurements method was applied. The experimental method precision is believed to be around 4.4% for Z-axis and 9.5% for X and Y axes, even though the experimental results validate the numerical method results with a relative error of 6.52% on average.