The deployment of a large number of loworbit constellations that will complete in the future has higher requirements for traditional space surveillance. The wideangle optical telescope system can monitor a wide area and observe more objects at the same time. However, most of the observation sequences are “short arc” observations, and a single sequence cannot determine the initial orbit of a space object. The current effective solution is to accurately correlate the optical observation arcs and fuse multiple observations for initial orbit determination. The orbits of two short arcs are optimized based on the admissible region method and the chisquare test is combined to determine the correlation between different observations. Secondly, the illposed problems are described for loworbit satellites in detail when anglesonly observations is used for correlation. Finally, given the high error rate of the loworbit object shortarc correlation, a method is proposed for identifying the wrong correlation based on the angular error characteristics.

To cope with the problem of the radar seeker measurement contains glint noise, interacting multiple model and robust filter application in the acceleration estimation of maneuvering target are studied. A novel algorithm named Highdegree Cubature Robust filter is proposed, in which HuberBased filter theory is used to develop Highdegree Cubature Kalman filter. Singer Model, Current Statistical Model, and Constant Acceleration Model are selected as mobile model of target. And the radar seeker measurement model is established. The target acceleration estimation filter is designed, which combines the High degree cubature robust filter with the interactive multiple model algorithm framework. The Monte Carlo simulation results show that the algorithm presented in this paper has a stronger robustness, and better accuracy in comparison with the Gaussian filter for the case of glint noise.

 The spacecraft attitude optimal control problem is considered, and the optimal attitude control problem is transformed into an optimal parameter selection problem by means of control parameterization and constrained transcription. A new interval time transformation method is proposed to solve the calculation of switching time for optimal parameter selection problem. The switching time is divided into a series of subsets in order, and the time domain width of each set is optimized as the decision variable. The switching time points in each subset change linearly with the decision variable. Thus, the control values distribute in several dense or sparse intervals, instead of being represented by evenly divided piecewise constant functions. Compared with the fulltime node optimization technique, this method effectively reduces the number of calculated decision variables and increases the execution efficiency of the algorithm.

The inherent random errors of gyroscopes accumulate more and more over time. Recurrent neural networks are widely used as an effective algorithm for processing time series signals. However, the traditional recurrent neural network (RNN) can not solve the longterm dependence in dealing with the random errors generated by the gyroscope, and it is prone to the problems of gradient disappearance and gradient explosion. In order to obtain accurate gyroscope signals, a denoising algorithm for gyroscope signals is proposed based on a variant of Long short term memory (LSTM) and gated recurrent unit (GRU).And the two networks are innovatively combined to verify. The random error of the original gyroscope is firstly analyzed through Allan variance, and then the output signal of the gyroscope is compensated based on the combination of LSTM and GRU. The results show that LSTM combined with GRU can significantly improve the random error processing of the gyroscope. X, Y, Zaxis gyroscope’s quantization noise, angle random walk, zerobias instability, angular velocity walk and speed ramp performance have been improved to varying degrees.