Optimal high-orbit Lambert transfer planning based on deep neural network
A fast solution scheme based on a deep neural network is proposed to solve the Lambert transfer planning problem for high-orbit on-orbit service and active space debris removal missions.Firstly,considering the J2 perturbation and coplanar circular orbit assumption,a representation of planning based on the Lambert transfer is established.Secondly,a two-step PSO algorithm considering the J2 perturbation is proposed for trajectory planning.The variation rules of △V cost relative to transfer time(△V-T curve characteristic)are summarized through simulation experiments,the sample form is defined,and the optimal fuel transfer knowledge base is constructed.Based on the summarized rules and the characteristics of the sample's nonlinear relationship,a fast trajectory planning strategy based on a deep neural network is proposed,which can reduce the number of calculating the Lambert transfer to two.Finally,simulation results demonstrate the effectiveness of the proposed optimal high-orbit Lambert transfer planning strategy.On the test set,the average absolute error of predicting time is 0.014 0％in the key area.The proposed strategy has wide application prospects.
deep neural networkon-orbit serviceLambert transferhigh-orbitfuel-optimal optimizationPSO
NETL
NSTL
万方数据
