Abstract:

The general dynamical model of satellite formation flying is derived. A sliding mode control method
based on neural networks is proposed to improve control accuracy and robustness of satellite formation flying. A
neural network based on radial basis function is designed to modify the parameters of exponent reaching law in
order to get an optimal balance between convergence speed of the sliding quantity and fuel consumption. Exponent
reaching law with saturation function is used to weaken chattering actuated by un-modeled dynamics and high
frequency switching control. The second order sliding quantity of the relative position error is used to improve the
control accuracy. Simulation results show effectiveness of the neural network-based sliding mode control method.

Key words: satellite formation flying, sliding mode control, neural network, chattering, robustness