Trajectory Optimization And Intelligent Control Of Asteroid Resource Exploitation And Utilization Mission

Resource exploitation and utilization of near-Earth asteroids provides not only rare and precious industrial metals,but also the sustainable energy for interplanetary travel,helping to take lead in inaugurating space economic circle.In recent years,a series of asteroid exploitation missions have been carried out.Missions such as asteroid hovering,asteroid landing,and sampling return from asteroid have been successfully conducted.However,asteroid mining missions are still full of challengs due to factors like fuel consumption,uncertain physical parameters of asteroids,irregular gravity,and delays in communication with the Earth.Accordingly,it is imperative to investigate methods of orbital design and motion control for the asteroid mining mission.This thesis focuses on optimization of transfer trajectory and studies on hovering and landing control in the vicinity of asteroids for asteroid mining missions.Low-thrust transfer has been widely used in deep space missions considering fuel consumption,particularly in asteroid missions.Indirect methods are easier to obtain a better solution than direct methods in optimization problems of low-thrust transfer trajectories.Nevertheless,indirect methods suffer from initializing co-state variables and convergence.In this paper,a dynamic model is established using logarithm of mass and acceleration as state variables instead of mass and thrust.The analytical expression of costate of logarithm of mass is derived,which makes co-state variables initialization much easizer.Secondly,time-optimal transfer and minimum thrust transfer is proved be equivalent under certain conditions.Time-optimal boundary equation is used to solve the minimum-thrust transfer orbit,which reduces the number of unknown co-states.Finally,a homotopy method from minimum-thrust trajectory to fuel-optimal trajectory is proposed,which improves the convergence of indirect method.In order to improve the efficiency of interplanetary transfer,variable-specificimpulse variable-thrust thrusters are becoming more popular in asteroid exploration missions.The fuel optimization model of variable-specific-impulse variable-thrust transfer trajectory is established under the constraints of thruster input power.The expression of optimal specific-impulse is obtained,and optimal specific-impulse and optimal power throttle level are decoupled.An exponential homotopy method of thruster input power is proposed,which solves the discontinuity of co-state equations.The obtained optimization result of the low-thrust transfer orbit is closer to engineering practice.The homotopy method connecting the double-specific-impulse trajectory and the variable-specific-impulse trajectory is proposed.Fuel consumptions of the doublespecific-impulse trajectory and the variable-specific-impulse trajectory are analyzed and compared.Asteroid hover control using reinforcement learning is studied for the sake of autonomy and robustness of spacecraft in uncertain asteroid environment.A reward function is proposed based on sliding mode control theory,which can lead the spacecraft to approach the hover target point.A dual-policy network is designed based on the reinforcement learning PPO algorithm for the sake of convergence of reinforcement learning training.The training process combines supervised learning and reinforcement learning to reduce training period.There is an increasing need to reduce the cost of asteroid mining and improve system reliability.Cubesat clusters are more suitable than large spacecrafts for high-risk missions near asteroids.In order to support the coordinated control and skill transfer between cubesats,a hierarchical reinforcement learning for cube satellite control is proposed.An option-PPO algorithm with fixed abstract time is proposed,which is a combination of subgoal-based algorithm with fixed abstract time and option-critic algorithm with variable abstract time.It simplifies the network structure of option-critic algorithm.Simulation results show that the option-PPO algorithm can accomplish obstacle avoidance and landing control in asteroid exploaration missions,and it is also suitable for asteroid hover control.