Estimation of aerodynamic parameters using Cascade Forward Back Propagation

In this paper, Cascade Forward Back Propagation (CFBP) algorithm based on artificial neural network is proposed for the estimation of longitudinal aerodynamic parameters of the aircraft. The proposed algorithm (CFBP) draws its inspiration from feed forward back propagation algorithm for estimating aerodynamic derivatives of longitudinal short period dynamics. The validation of CFBP algorithm is done on simulated flight data of research aircraft `HANSA' using various combinations of elevator inputs. The results obtained using the proposed algorithm is also compared with parameters estimated using conventional Feed Forward Back Propagation (FFPB) algorithm. As compared to FFBP algorithm, CFBP algorithm yields estimates with lesser standard deviation with best training performance in less number of iterations. The results suggest that the CFBP algorithm can be used advantageously to develop relationship between motion/control variables and aerodynamic coefficients and for the estimation of aircraft parameters.