For a class of cube satellites using 3-axis magnetorquers in low earth orbit,an adaptive tracking control method is presented based on weighted multiple-model method,in order to make the attitude control system track the attitude variation within the tolerant error range quickly and precisely.First,mathematic models of attitude control system at different equilibriums are obtained to form multiple-model structure which can be controlled by adjusting weights online based on model-matching performance index.Next,a dynamic adaptive neural network is embedded into the multiple-model control loop to eliminate the influence of earth magnetic field variation in-orbit operation,and also deal with uncertain external disturbances to improve the robust control performance,due to the good approximating ability of dynamic neural network for uncertain nonlinear function.Finally,the numerical simulation is performed for a certain cube satellite,and the results show that the control method presented is effective.