On-orbit Servicing Trajectory Planning Method Under Incomplete Information Conditions

Aiming at the issue of parameter uncertainty in on-orbit servicing tasks，a trajectory planning method based on dynamic planning period adjustment and short-term repetitive cycle-induced optimization using neural networks is proposed. Through the real-time learning and prediction of temporal information by neural networks，the period of trajectory planning is dynamically adjusted to achieve a balance between computational efficiency and planning accuracy. By integrating receding horizon optimization and asynchronous optimization strategies， trajectory parameters are repetitively refined within a local time window. High-frequency short-term local optimization is employed to suppress cumulative errors，thereby ensuring the stable operation of spacecraft in complex space environments and the successful accomplishment of missions. Simulation results demonstrate that，under conditions of incomplete information，the proposed method can control the trajectory of spacecraft more effectively，enabling them to maintain stability in complex space environments. Monte Carlo simulation results reveal that the proposed method achieves a success rate of 81. 34%. In the majority of samples，the algorithm is capable of handling the corresponding incomplete information and resolving the trajectory planning problem，showcasing high reliability and robustness.