Experimental design and analysis techniques featuring a factorial design were applied to a popular multi-rotor flight simulation tool, eCalc, to validate its hover time flight prediction capability for a quadcopter style unmanned aerial vehicle (UAV). The general applicability of formal experiment design was demonstrated for unmanned aircraft system flight test.
A representative quadcopter was chosen to illustrate the efficacy of the statistically based flight test methodology. Five factors including battery capacity, battery cell count, propeller diameter, propeller pitch and video transmitter power were used in a 25-1 fractional factorial design. All factors were presumed to have an influence on the flight profile. Flight data from 16 separate flights were gathered, using a fully randomized test schedule. Similarly, identical multi-rotor configurations were evaluated in eCalc using a full factorial 25 design.
Regression models from each experiment demonstrated a strong model where all five factors and many interactions were significant. A total of 16 flight data test points and five confirmation points showed hover time predictions were within ± 15% accuracy as eCalc claims. Results validated eCalc as a reliable model for predicting multi-rotor hover times. The fractional factorial proved an effective tool for efficient flight test, an inherently low signal/noise environment.
Formal experiment design applied to flight test is rarely featured in the open literature. This study highlights the reliability of the method and eCalc as a prediction tool for quadcopter hover times, adding to its credibility for quadcopter style UAV flight planning and design.
