Modelling missile motion system using neural networks

摘要

The models of nonlinear systems are identified by recursive predictive errors (RPE) method based on the layered neural networks. To improve the identification precision, gain coefficient and momentum factor are introduced into the algorithm for the data are disturbed by noises and vary suddenly. This algorithm is applied to the dynamic modelling of rolling and pitching angles of missiles. Simulation results show that the proposed algorithm is suitable for the modelling of nonlinear systems.