Nonlinear multi-input multi-output system identification using neuro-evolutionary methods for a quadcopter

This research focuses on studying the effect of using evolutionary algorithms in improving neural network capabilities in identification of non-linear multi-input and multi-output dynamic systems such as a quadcopter. In addition, comparison of the different neural network based approaches is carried out in order to reveal the variations among the different methods. The results show that using evolutionary algorithms in training a neural network enhanced the system identification accuracy. Furthermore, the results show that differential evolution neural networks have promising potential to be used in multi-input multi-output system identification.