Particle Swarm Optimization Applied to Spacecraft Reentry Trajectory
Particle Swarm Optimization Applied to Spacecraft Reentry Trajectory
复制标题
DOI：
10.2514/1.56387
复制
发表时间：
2013-01
影响因子：
2.6
通讯作者：
Afshin Rahimi;Krishna Dev Kumar;H. Alighanbari
中科院分区：
文献类型：
--
作者：
Afshin Rahimi;Krishna Dev Kumar;H. Alighanbari
非线性最优控制问题的数值解通常没有解析解，可以根据其各自的优点和特点分为不同的类别。在直接方法中，所考虑系统的模型方程被离散化，并且控制轨迹被参数化以获得有限维参数优化问题[1]。在这些直接方法中，全局优化方法（也称为进化算法）近年来已经变得令人感兴趣，并且在该领域已经进行了各种研究[1-4]。这一类别中的一些众所周知的方法是遗传算法（GA），其基于达尔文的适者生存原则对物种的进化进行建模;模拟退火（SA），其在退火过程中模拟大量原子的平衡;以及蚁群优化，其受到蚂蚁行为的启发。在所有的全局优化技术中，基于群体智能（SI）的方法由于其速度和精度而变得越来越流行。它们的灵感来自自然现象，如鸟类，蚁群，动物群的行为，甚至人类之间的社会联系[3]。粒子群优化（PSO）的思想是在1995年由Eberhart和Kennedy [5]首次提出的，然后由其他研究人员[6]进行了修改。SI方法中最重要的因素是，因为它们使用了一组个体的整个经验，而不仅仅是每个个体粒子的经验（即，一个潜在的解决方案），他们的收敛速度比其他方法更快。本说明的范围是提出一种使用PSO方法解决最优控制问题的新方法，并避免常见分析方法中所需的计算。这是通过使用特定问题的现有解决方案，然后尝试为其他感兴趣的目标找到其他可能的轨迹来实现的。本说明的组织如下：在第二节。第二，在[7]的基础上建立了再入航天器的运动方程组。节中介绍了粒子群优化方法和映射过程。节中IV的结果是两个不同的目标函数。第一个目标是验证所提出的方法的准确性和其他目标是尽量减少积分的热量施加到航天器。
THE numerical solution of optimal control problems that are nonlinear and, therefore, generally without analytical solutions, can be categorized into different classes with their own advantages and characteristics. In the direct approach, the model equations of the considered system are discretized, and the control trajectories are parametrized to obtain a finite-dimensional parameter optimization problem [1]. Among these direct methods, global optimization methods (also known as evolutionary algorithms), have become of interest in recent years and various research has been done in this area [1–4]. Some well-known methods in this category are genetic algorithms (GAs), which model the evolution of species based on Darwin’s principle of survival of the fittest; simulated annealing (SA), which mimics the equilibrium of large numbers of atoms during an annealing process; and ant-colony optimization, which is inspired by the behavior of the ants. Among all global optimization techniques, the swarm intelligence (SI)-based methods are becoming more popular due to their speed and accuracy qualities. They are inspired by natural phenomena such as the behavior of groups of birds, ant colonies, herds of animals, and even social connections between human beings [3]. The idea of particle swarm optimization (PSO) that is addressed in this Note was first introduced in 1995 by Eberhart and Kennedy [5] and was then followed and modified by other researchers [6]. The most important factor that stands out in SI methods is that because they use the whole experience of the group of individuals, rather than only the experience of each individual particle (i.e., one potential solution), their convergence speed is faster than other methods. The scope of this Note is to present a new method for solving an optimal control problem using a PSO method and avoiding the calculations needed in the common analytical approaches. This is accomplished by using an existing solution for a specific problem and then trying to find other possible trajectories for other objectives of interest. This Note is organized as follows: in Sec. II, the system of equations ofmotion for a reentry spacecraft is presented based on [7]. In Sec. III the PSO optimization method and mapping procedure are described. In Sec. IV the results are presented for two different objective functions. The first objective is for validating the proposed method in terms of accuracy and the other objective isminimizing the integral of the heat applied to the spacecraft.