Abstract
System identification provides a process to develop different dynamic models of varying structures based on user-defined requirements. For a quadrotor, system identification has been primarily in the field of off-white and grey-box models, but black-box models have the advantage of incorporating nonlinear aero-dynamic effects while also maintaining performance. For the identification, both a chirp and Hebert-Mackin parameter identification method waveform are used as inputs to maximize excitation while minimizing nonlinear responses. The quadrotor structure is defined by the an autoregressive with exogenous input (ARX) model, an autoregressive-moving-average (ARMAX) model, and a Box-Jenkins (BJ) models and then identified with the prediction error method. The black-box method shows that it maintains identification performance while improving upon the flexibility of different cases and ease of implementation.
