基于遗传编程的航天器有限推力逼近轨迹规划

通过引入基函数的概念,提出了采用遗传编程求解有限推力航天器逼近非合作目标最终逼近段轨迹规划问题的方法。该方法将推力器开关状态定义为基函数,以多个基函数分别乘以开关状态持续时间再求和作为推力器开关的历程函数;将历程函数转换为遗传编程的树型结构,将消耗燃料的质量作为适应度函数,并将规避障碍物和终端逼近精度等约束条件以罚函数的形式添加到适应度函数中;利用遗传编程的模拟自然进化理论的全局寻优机制求解,最终得到最优逼近轨迹方案。某航天器在有限推力下逼近非合作目标的轨迹规划结果表明:整个逼近过程推力器仅开关5次,大大降低了对开关频率的要求,同时,规划结果比采用高斯伪谱法时逼近时间降低了30.09%,燃料消耗降低了4.18%。

Finite thrust spacecraft approaching trajectory planning based on genetic programming

Based on the genetic programming algorithm with the introduction of the basis function, this paper proposes a new way to deal with finite-thrust spacecraft trajectory optimization when approaching a non-cooperative object in the final approach phase. Thruster switch state is defined as the basis functions which respectively multiply by the switch state duration and then sum as the thruster state function. The thruster state function is converted into genetic programming tree structure. Fuel consumption is defined as the fitness function of genetic programming, and the constraint conditions of obstacle avoidance and approaching precision are induced into the fitness function in the form of penalty function. Genetic programming algorithm is used to obtain the optimal-fuel trajectory planning scheme by training all the possible combinations of thrusters switch state. The result is global optimal, and the thrusters switch frequency is not high. The trajectory planning result based on finite thrust of a spacecraft approaching non-cooperative object shows that the thruster is only opened and closed 5 times, greatly reducing the requirement for switching frequency. Approaching time is reduced by 30.09% compared with the Gauss pseudospectral method, and fuel consumption is reduced by 4.18%.