On-orbit Estimation Of Unbalanced Torques Acting On Satellite Platform With Large Inertia Rotating Payloads

With the increasing requirements of space missions undertaken by satellites,threeaxis stabilized satellites that are in steady-state operation are often equipped with various types of rotating payloads.Considering a kind of satellites with large inertia rotating payloads,in order to achieve a high precision attitude control,starting from studying the unbalanced characteristics of large inertia rotating payload,the coupling relationship between satellite platform and large inertia rotating payload is analyzed.Furthermore,it is necessary to adopt effective estimation algorithms for on-orbit identification of unbalanced disturbance torques caused by the rotating payloads.The main contents of this paper are as follows:Firstly,the dynamic analysis and modeling of the large inertia rotating payloads are carried out.The nature of static and dynamic imbalance is quantitatively analyzed,and the main influencing factors of the unbalanced torque are determined.Considering the condition that the rotating payload rotates at a constant speed,the unbalanced moment when the rotating payload rotates is regarded as the external disturbance,and the simplified satellite attitude dynamic equation is provided.Furthermore,the multi-rigid pose dynamics modeling method is used to derive the attitude dynamic equation of the satellite considering the coupling effect.By comparing the unbalanced disturbance torques based on the two models through mathematical simulation,it can be seen that the dynamics model of satellites considering the effects of coupling is more representative of the dynamic characteristics of such kind of satellites.Considering that the attitude of the satellite remains stable,on the basis of the approximate linearization of the state equation of the satellite dynamic,the unbalanced disturbance torque of the rotating payload is regarded as an unknown input vector added to the satellite platform.Linear estimation methods such as unknown input observer based on nonlinear tracking differentiator is used.By adding a nonlinear tracking differentiator to the unknown input observer instead of direct derivation of the satellite attitude information,the performance of the unknown input observer is improved.The influence of the measurement noise of the sensor on the estimation result is suppressed.The dynamics of the satellite are nonlinear under the satellite attitude maneuvering condition.Based on the Two-Stage Kalman filter,the Two-Stage Extended Kalman Filter(TSEKF)is derived by combining the extended Kalman filter which is used in nonlinear systems.It is suitable for the estimation of the unbalanced torque of the rotating payload under the satellite attitude maneuvering condition.Through mathematical simulation,the validity of the TSEKF for estimating the unbalanced torque is verified.And TSEKF has the advantages of small computation and accurate estimation.Under the same noise condition,the Two-Stage Extended Kalman filter is more accurate in estimating the amplitude of the unbalanced torque by comparing with the estimation results of the unknown input observer based on the nonlinear tracking differentiator.Considering that satellite dynamics is approximated based on the aforementioned methods,in the case of completely retaining the nonlinearity of the satellite attitude dynamics,through the training of a large amount of data,a Deep Feedforward Network(DFN)is used to extract the characteristic relationship between the unbalanced torques of the rotating payload and the angular velocity of rotating payload,the output torques of the satellite actuator as well as the measurement information of the sensors.Furthermore,a Deep Feedforward Network model with the capable of fitting satellite dynamics in high precision is obtained.The simulation results show that the trained Deep Feedforward Network can quickly and accurately estimate the unbalanced torque when the attitude of the satellite is stable.