Recursive Gauss-Newton based training algorithm for neural network modelling of an unmanned helicopter dynamics

This paper presents a recursive Gauss-Newton based training algorithm to model the dynamics of a small scale helicopter system using neural network modelling approach. It focuses on selection of optimized network for recursive algorithm that offers good generalization performance with the aid of the cross validation method proposed. The recursive method is then compared with off-line Levenberg-Marquardt (LM) training method to evaluate the generalization performance and adaptability of the model prediction. The results indicate that the recursive Gauss-Newton method gives slightly lower generalization performance compared to its off-line counterpart but adapts well to the dynamic changes that occur during flight. The proposed recursive algorithm was found effective in representing coupled helicopter dynamics with acceptable accuracy within the available computational timing constraint.