Uncertainty Quantification And Robust Optimal Guidance Methods For Mars Entry

Mars entry trajectory planning and guidance is one of the key technologies for Mars landing mission.However,the initial state uncertainty and dynamics parameter uncertainty in Mars entry process have become a bottleneck restricting the further accuracy improvement of the guidance for Mars entry.Uncertainty quantification and robust optimal guidance problems for Mars entry remain to be solved urgently.In order to reduce the risk and improve the guidance accuracy for Mars entry,this dissertation focuses on uncertainty quantification,robust trajectory optimization,and robust optimal guidance,while a set of theoretical methodology and algorithms on Mars entry uncertainty quantification and guidance planning are developed.First,two categories of methods on uncertainty quantification for Mars entry are proposed.In the framework of intrusive polynomial chaos,an uncertainty quantification method for Mars entry based on adaptive generalized polynomial chaos is proposed.The dramatic increase of the number of equivalent deterministic differential equations in solving high-dimensional uncertainty quantification problem is depressed through stochastic space decomposition.The stochastic space basis functions are adaptively updated by spectral decomposition to improve the solution accuracy of uncertainty quantification under the conditions of highly nonlinear dynamics and high-dimensional uncertainty.In the framework of non-intrusive polynomial chaos,an uncertainty quantification method for Mars entry using sensitivity collocation based non-intrusive polynomial chaos is proposed.By evaluating the sensitivity of uncertainty and combining with Latin hypercube collocation,a more appropriate statistical description for mixed uncertainty with different distribution types is achieved,thus ensuring the efficiency and accuracy of mixed uncertainty quantification.The presented two methods are applied to reveal the basic laws of quantification and propagation of the initial state uncertainty and dynamics parameter uncertainty as well as their coupling uncertainties for Mars entry.Second,two types of methods on robust optimization for Mars entry trajectories are presented.In order to rapidly plan the Mars entry trajectory under uncertainty,a hybrid optimization algorithm based on particle swarm optimization and Gauss pseudospectral method is developed.Through successive optimization combining global and local optimization operations,the efficiency of optimal trajectory generation for Mars entry is improved.In order to obtain the optimal reference trajectory with robustness and reliability,uncertainty quantification based robust optimization method for Mars entry trajectory is proposed.By incorporating the effects of initial state uncertainty and dynamics parameter uncertainty on dynamics,objective function and constraints into the Mars entry trajectory optimization process,the generated trajectories are enabled to synchronously reach the optimality and robustness of the performance index and the reliability of constraints satisfaction.Third,two kinds of algorithms of robust optimal guidance for Mars entry are proposed.In order to track the robust optimal nominal trajectory accurately under uncertainty,radial basis function neural network is employed to approximate the uncertainty elements in tracking error,and a robust tracking guidance algorithm for Mars entry based on second-order sliding mode and radial basis function neural network is developed.To enhance the robustness of optimal guidance under uncertainty,a collaborative optimal guidance approach based on reinforcement learning and pseudospectral method is proposed.A collaborative optimization framework is constructed,and the optimal Mars entry guidance and optimal power descent guidance are coordinated by solving optimal handover online,thus the integrated guidance under uncertainty reaches both robustness and optimality.