Spacecraft's Automatic Landing Control based on online tracing identification method of neural network

摘要

To improve the dynamical property and decoupling capability for a class of spacecraft system with strong coupling, a neuron-network controller based on online tracing identification is established to meet the decoupling requirements of multivariable system. The model with new structure and learning algorithm has significance for weight matrices and makes training process of weights become more distinct and straightforward. The new neural network is then applied to identification of nonlinear dynamics system, which the speed of learning and convergence is improved greatly for using the priori input-output state knowledge. The results of simulation show that the neuron network decoupling controller based on online tracing identification can effectively reduce the identification errors caused by the different sampling data, and improve prominently the precision and the reliability of neural network in the system identification. The controller has powerful self-learning and self-adaptive decouple capabilities.