Solar sail trajectory optimization using rapidly converging evolutionary neural network

Abstract

Solar sail is a propulsion system of spacecraft that uses solar wind on a large area of sail. Recently many researches have been done because it is considered to appropriate for long-term space mission such as interplanetary mission due to its property that there is no fuel consumption and few mechanical driving part. But it is difficult to control the magnitude of thrust and the direction of thrust independently, and the magni-tude of thrust itself is so low that finding optimal control is a very important problem for designing a new so-lar sail mission. Generally the trajectory optimization problem for continuous thrust is known to be a hard problem and consumes large computational time. Among many of studies to solve that problem, the existing method using evolutionary neural network has a strong point that it is able to find global optimum solution without any initial guess. In this research, we implement the existing evolutionary neurocontroller and find the optimal shape of it by solving trajectory optimization problem. But when we use genetic algorithm, it shows poor conver-gence speed and poor accuracy on the solution caused by the property of GA itself. In other to solve that problem, a new approach for neurocontroller that utilize other EAs is introduced. Particle swarm optimization, Memetic algorithm, and also their hybridization algorithm was proposed to improve the performance of evo-lutionary neurocontroller. The way to applying these EAs on neural network individual is studied. We implement the proposed algorithms and apply them to the trajectory optimization problem for solar sail Earth-Moon transfer. The results and the performance of proposed algorithms are compared with the ones of existing evolutionary neurocontroller. Additionally we have simulation for many other cases with different efficiency of solar sail in other to confirm the practical ability of proposed algorithm.