人工智能在航天器制导与控制中的应用综述

航天器制导与控制技术是保障空间任务顺利实施的关键技术之一。当前,动力学模型的强非线性以及参数不确定性制约了高精度姿轨控技术的发展,而系统故障则决定航天器姿轨控的成败。以机器学习为代表的新一代人工智能技术航天器制导控制领域展现了巨大的应用潜力。首先对基于人工智能技术的轨迹制导和姿态控制中的研究发展及应用现状进行归纳,分析航天器轨迹规划、姿态控制、故障诊断以及容错控制技术的发展趋势。然后,从鲁棒轨迹规划、自适应姿态控制、快速故障诊断和自适应容错控制等4个方面总结适用于未来航天任务的航天器姿轨控关键技术。最后,针对智能姿轨控技术的应用所面临的挑战,从姿轨控架构、算法最优性、算法的训练以及技术验证等方面提出相应的发展建议。     

Impact of mission requirements and constraints on conceptual launch vehicle design

The objective of this paper is to analyse the impact of mission requirements and constraints on both the optimum vehicle design and the effects on flight path selection for two types of reusable two-stage-to-orbit launch vehicles. The first vehicle type considered provides horizontal take-off and landing capabilities and is intended to be propelled by an airbreathing propulsion system during stage 1 flight. The second vehicle type assumes a vertical launch and is accelerated by a rocket propulsion system during the booster stage ascent flight. The analysis employs a design tool for simultaneous system and mission optimization. It consists of a CAD-based preliminary vehicle design tool, aerodynamic and aerothermodynamic calculation software, flight simulation programs, and a two-level decomposition optimization algorithm enabling simultaneous system and flight optimization. The results to be presented show that the cruise flight requirement for an European launched mission of the airbreathing vehicle results in a loss of 60 % payload mass as compared to a mere accelerated ascent for a near equatorial mission into the same target orbit assuming constant take-off mass. The strong dependencies of mission requirements on both the optimal vehicle design and the ascent performance are determined for the rocket-powered vehicle type by varying the inclination and altitude of the target orbit.     

可重复使用飞行器导航制导控制技术展望

针对当前可重复使用飞行器的发展现状,总结了其特点、优势和发展趋势.通过飞行任务,分析了可重复使用飞行器所面临的复杂环境特性和飞行控制特性.从飞行控制角度出发,概括了可重复使用飞行器导航制导控制所面临的技术挑战,包括高精度导航技术、多约束航迹优化技术、再入段和能量管理段的制导技术以及强耦合强不确定强鲁棒姿态控制技术等.最后,针对技术挑战,对可重复使用飞行器导航制导控制技术提出了若干建议.     

考虑控制系统故障的重复使用运载器轨迹优化

传统的重复使用运载器的制导系统与控制系统的设计是分开进行的。当运载器出现故障时,需要进行轨迹和控制的重构,故障下气动特性和配平能力的变化使得重构后的制导系统与控制系统难以良好匹配。考虑到控制系统对轨迹设计和优化的影响,通过将三自由度轨迹优化与多舵面控制分配结合,提出一种考虑控制舵面气动力和故障的重复使用运载器轨迹优化方法。仿真结果表明,多舵面控制分配算法可以将运载器故障下的配平能力转化为轨迹优化的约束条件,从而避免了传统三自由度轨迹优化方法结果无法被控制系统跟踪的可能性,并为制导与控制一体化设计提供参考依据。     

交会对接任务轨道控制规划设计与实施简

针对我国空间交会对接轨道控制规划技术,研究了轨道交会优化、应急轨道控制、安全轨道防护和发射窗口规划等一系列关键问题.设计了全寿命周期交会对接任务轨道控制规划方案,从目标飞行器发射到飞船返回,对轨道控制进行了全程协同、全局优化.设计了相位、高度、圆化度多目标融合控制算法;建立了规划变量对远距离导引终点六自由度的独立控制方程;设计了标称整体规划与动态逐级规划相结合的多模式规划策略;基于导引终点整体调整和局部调整的方式,实现了正常和应急条件下天地导引交接点的动态规划;提出了基于飞行控制过程建模的导引终点精度分析方法,确定了地面导引向自主导引切换的关键判据;建立了多约束交会对接发射窗口模型,构建了多任务多年度发射窗口集合.交会对接轨道控制规划技术成功应用于神舟八号、神舟九号和神舟十号交会对接任务.     

大跨时空任务背景下的太阳能无人机任务规划技术研究进展

通过任务规划技术合理的优化太阳能无人机的飞行轨迹和动力学参数，能够有效提高太阳能无人机的能量利用率，使其胜任许多大范围跨时间跨空间飞行任务。从能量建模、续航评估和能量管理策略3个方面对大跨时空任务背景下太阳能无人机任务规划技术的研究进展进行了综述。在能量建模方面，介绍了当前主流的太阳辐射模型和能量生产基本框架；在续航评估方面，分析了目前的指标设计和应用方法；在能量管理策略方面，从能量综合应用、风力滑翔机制、轨迹优化方法和面向特定任务的应用4个角度，梳理了当前的研究现状。最后，对该领域未来可能的研究方向进行了展望。     

基于遗传算法的无人机协同侦察航路规划

无人机将成为侦察卫星、有人驾驶侦察机的重要补充与增强手段 ,成为未来战场上广泛应用的一种侦察工具。为了提高无人机 (UAV)的侦察效率 ,在执行侦察任务前必需规划设计出高效的无人机侦察飞行航路。针对这一问题 ,本文提出了一种侦察效率指标评估的计算方法 ,解决了航路规划中的侦察效率量化问题。考虑在大范围任务区域内进行侦察航路优化存在计算的复杂性和收敛性等问题 ,本文采用遗传算法对侦察航路进行了优化处理。通过该方法得到的侦察航路可以有效地提高无人机的侦察效率。     

基于NFTET的高超声速飞行器再入容错制导

针对以X-33为对象的三自由度高超声速飞行器,采用相邻可行轨迹存在定理(NFTET)设计了容错制导律以解决再入段执行器发生故障的轨迹重构问题。在标称情况下采用预测校正算法生成满足再入过程约束和终端约束要求的再入轨迹;当执行器发生故障时,飞行器气动参数、结构和舵面力矩都可能发生不可预测的变化,原先的轨迹不再满足制导要求,因此需要设计新型容错制导律。针对实际再入制导模型,基于NFTET设计容错制导算法对轨迹进行重构,得到满足故障情况下制导任务的可行轨迹。从仿真结果中可以看出,容错制导算法生成的新轨迹重新回到了约束范围之内,轨迹呈收敛趋势,使得高超声速飞行器从故障恢复到正常飞行状态,提高了飞行器的自主容错能力。     

Flight trajectory optimization of sun-tracking solar aircraft under the constraint of mission region

The optimal yawing angle of sun-tracking solar aircraft is tightly related to the solar azimuth angle, which results in a large arc flight path to dynamically track the sun position. However, the limited detection range of payload usually requires solar aircraft to loiter over areas of interest for persistent surveillance missions. The large arc sun-tracking flight may cause the target area on the ground to be outside the maximum coverage area of payload. The present study therefore develops an optimal flight control approach for planning the flight path of sun-tracking solar aircraft within a mission region. The proposed method enables sun-tracking solar aircraft to maintain the optimal yawing angle most of the time during daylight flight, except when the aircraft reverses its direction by turning flight. For a circular region with a mission radius of 50 km, the optimal flight trajectory and controls of an example Λ-shaped sun-tracking solar aircraft are investigated theoretically. Results demonstrate the effectiveness of the proposed approach to optimize the flight path of the sun-tracking aircraft under the given circular region while maximizing the battery input power. Furthermore, the effects of varying the mission radius on energy performance are explored numerically. It has been proved that both net energy and energy balance remain nearly constant as the radius constraint varies, which enables the solar aircraft to achieve perpetual flight at almost the same latitude as the large arc flight. The method and results presented in this paper can provide reference for the persistent operation of sun-tracking solar aircraft within specific mission areas.     

基于二次点火的飞行器横向转移入轨弹道设计

为了解决航天器发射过程因发射场地的地理位置约束造成的入轨夹角问题,基于火箭上面级二次点火,对航天器横向转移零速度偏差入轨弹道进行了设计。对二次点火和变射面横向转移弹道进行了研究,根据横向转移弹道特点,为简化控制程序,分时序对火箭二级非连续点火纵横向飞行程序进行了设计;并分析横向转移入轨弹道各项约束条件,建立弹道优化模型。根据弹道设计模型,以某两级液体燃料运载火箭为研究对象,对二次点火横向转移入轨弹道进行优化仿真。结果表明:入轨时刻航天器位置偏差为0.391m,速度偏差为1.277m/s,速度偏差比二次点火固定射面入轨弹道减少了737.844m/s,满足零速度偏差入轨精度要求。      

Space Flight Validation of Design and Engineering of the ZDPS-1A Pico-satellite

The ZDPS-1A pico-satellites are the first satellites in China within the 1-10 kg mass range that are successfully operated on orbit. Unlike common pico-satellites, they are designed to be "larger but stronger" with more powerful platforms and unique payloads so as to bear a better promise for real applications. Through their space flight mission, the functionality and performance of the two flight models are tested on orbit and validated to be mostly normal and in consistency with design and ground tests with only several inconforming occasions. Moreover, they have worked properly on orbit for one year so far, well exceeding their life expectancy of three months. Therefore, the space flight mission has reached all its goals, and verified that the design concept and the engineering process of the picosatellites are sufficient in allowing them the desired functionality and performance in, and the adaption to the launch procedure and the low-Earth orbit space environment. In the foreseeable future, the platform together with the design concept and the engineering process of the pico-satellites are expected to be applied to more complicated real space applications.     

小推力深空探测轨道全局优化设计

 针对小推力深空探测四维轨道优化设计,给出了一种组合优化算法,采用该算法基于二体模型进行了深空探测四维轨道全局优化。该组合优化算法由动态规划算法、静态参数优化算法与共轭梯度算法组成。动态规划算法和静态参数优化算法用以选择最优的发射窗口、返回窗口及相应的近似飞行轨道;基于该近似轨道方案,采用共轭梯度算法（解决两点边值问题）求解精确的最优轨道。通过大量的数值仿真计算,得到了航天器的全局最优飞行轨道,及相应的最优发射窗口与返回窗口。数值仿真结果表明,该组合优化算法对深空探测轨道优化具有良好的通用性和工程运用价值。     

Trajectory optimization of multiple quad-rotor UAVs in collaborative assembling task

A hierarchic optimization strategy based on the offline path planning process and online trajectory planning process is presented to solve the trajectory optimization problem of multiple quad-rotor unmanned aerial vehicles in the collaborative assembling task. Firstly, the path planning process is solved by a novel parallel intelligent optimization algorithm, the central force optimization-genetic algorithm (CFO-GA), which combines the central force optimization (CFO) algorithm with the genetic algorithm (GA). Because of the immaturity of the CFO, the convergence analysis of the CFO is completed by the stability theory of the linear time-variant discrete-time sys-tems. The results show that the parallel CFO-GA algorithm converges faster than the parallel CFO and the central force optimization-sequential quadratic programming (CFO-SQP) algorithm. Then, the trajectory planning problem is established based on the path planning results. In order to limit the range of the attitude angle and guarantee the flight stability, the optimized object is changed from the ordinary six-degree-of-freedom rigid-body dynamic model to the dynamic model with an inner-loop attitude controller. The results show that the trajectory planning process can be solved by the mature SQP algorithm easily. Finally, the discussion and analysis of the real-time per-formance of the hierarchic optimization strategy are presented around the group number of the waypoints and the equal interval time.     

New Horizons Mission Design

In the first mission to Pluto, the New Horizons spacecraft was launched on January 19, 2006, and flew by Jupiter on February 28, 2007, gaining a significant speed boost from Jupiter’s gravity assist. After a 9.5-year journey, the spacecraft will encounter Pluto on July 14, 2015, followed by an extended mission to the Kuiper Belt objects for the first time. The mission design for New Horizons went through more than five years of numerous revisions and updates, as various mission scenarios regarding routes to Pluto and launch opportunities were investigated in order to meet the New Horizons mission’s objectives, requirements, and goals. Great efforts have been made to optimize the mission design under various constraints in each of the key aspects, including launch window, interplanetary trajectory, Jupiter gravity-assist flyby, Pluto–Charon encounter with science measurement requirements, and extended mission to the Kuiper Belt and beyond. Favorable encounter geometry, flyby trajectory, and arrival time for the Pluto–Charon encounter were found in the baseline design to enable all of the desired science measurements for the mission. The New Horizons mission trajectory was designed as a ballistic flight from Earth to Pluto, and all energy and the associated orbit state required for arriving at Pluto at the desired time and encounter geometry were computed and specified in the launch targets. The spacecraft’s flight thus far has been extremely efficient, with the actual trajectory error correction ΔV being much less than the budgeted amount.     

基于混合优化的运载器大气层内上升段轨迹快速规划方法

针对运载器大气层内的最优轨迹快速规划问题,提出一种将求解最优控制问题的间接法与直接法相结合的混合优化方法。首先,基于最优控制问题的一阶必要条件,将运载器大气层内的三维最优上升问题转化为Hamiltonian两点边值问题;然后,采用直接法中能以较少的节点获得较高求解精度的Gauss伪谱法进行求解,提高算法的求解效率;最后,采用真空解析解初值及密度同伦技术,解决初值猜测与算法收敛困难的问题。仿真结果表明,混合优化算法能够准确、快速地对运载器大气层内的最优上升轨迹问题进行求解,并在计算精度与效率上均优于间接法,可应用于运载器的轨迹在线规划与闭环制导。     

Micromission spacecraft: a low-cost, high-capability platform for space science missions

The Ball Micromission Spacecraft (MSC) is a multi-purpose platform capable of supporting science missions at distances from the Sun ranging from 0.7 to 1.7 AU. In the baseline scenario, MSC is launched as a secondary payload on an Ariane 5 rocket from Kourou, French Guiana, to GTO using the Ariane 5 structure for auxiliary payloads (ASAP5). The maximum launch wet mass is 242 Kg and can include up to 45 Kg of payload depending on AV needs. The on-board propulsion system is used for maneuvering in the Earth-Moon system and injecting the spacecraft into its final orbit or trajectory. For Mars missions, MSC enables orbiting Mars for science payloads and/or communications and navigation assets, or for precision Mars fly-bys to drop up to six probes. The micromissions spacecraft bus can be used for science targets other than Mars, including the Moon, Earth, Venus, Earth-Sun Lagrange points, or other small bodies. This paper summarizes the current spacecraft concept and describes the multimission spacecraft bus implementation in more detail.     

人工智能在航天飞行任务规划中的应用研究

 在航天飞行任务中，如何设计航天器的飞行过程，如何确定地面对航天器的控制操作，如何制定飞行控制计划等，是地面飞行控制中心面临的重大问题，也是航天飞行任务规划所要解决的基本问题。在充分认识和把握人工智能基本原理、方法和技术的基础上，提出了一个基于规则演绎和状态演化的生长式推理模型，并对模型的特性进行了详细讨论，然后导出了该模型在航天飞行任务规划问题中的具体形式，从而成功地解决了航天任务自动规划的难题。通过在实际航天任务中的应用和验证，不仅证明该模型和方法是正确的、可行的和高效的，而且证明人工智能在航天飞行任务规划中有着广阔的应用前景。     

Integrated Entry Guidance for Reusable Launch Vehicle

A method for the implementation of integrated three-degree-of-freedom constrained entry guidance for reusable launch vehicle is presented. Given any feasible entry conditions, terminal area energy management interface conditions, and the reference trajectory generated onboard then, the method can generate a longitudinal guidance profile rapidly, featuring linear quadratic regular method and a proportional-integral-derivative tracking law with time-varying gains, which satisfies all the entry corridor constraints and meets the requirements with high precision. Afterwards, by utilizing special features of crossrange parameter, establishing bank-reversal corridor, and determining bank-reversals according to dynamically adjusted method, the algorithm enables the lateral entry guidance system to fly a wide range of missions and provides reliable and good performance in the presence of significant aerodynamic modeling uncertainty. Fast trajectory guidance profiles and simulations with a reusable launch vehicle model for various missions and aerodynamic uncertain-ties are presented to demonstrate the capacity and reliability of this method.     

基于混合粒子群算法的上升段交会弹道快速优化设计

基于梯度搜索的高效性和粒子群搜索的随机性,提出了一种混合粒子群算法,并应用该算法研究了运载火箭上升段交会弹道快速优化设计问题.以运载火箭与目标飞行器在交会时刻的距离最小为目标函数,设计了运载火箭飞行程序,建立了运载火箭上升段交会弹道优化模型,同时分别采用混合粒子群算法、遗传算法和粒子群算法进行求解.仿真结果表明:基于本文算法对运载火箭上升段交会弹道进行优化设计,平均交会位置误差为4.137m,较遗传算法减少了17.940m,平均优化耗时488.922s,较粒子群算法缩短了2342.125s.混合粒子群算法搜索速度较快,收敛精度较高,可用于运载火箭上升段交会弹道的快速优化设计.      

基于地面任务-空间姿态映射的敏捷卫星任务规划

面向观测时间窗口相互重叠的多点目标观测任务需求,对敏捷卫星单星单轨任务规划问题进行研究。针对传统方法在卫星机动能力受限和成像任务冗余两种情况下求解效率低的缺陷,引入任务-姿态协同规划思想。首先,建立地面任务和空间姿态映射关系,并考虑相邻任务间姿态机动时间的最优性使得卫星在观测相邻任务时无多余等待时间,以此来设计任务-姿态协同规划数学模型。其次,根据任务-姿态协同规划数学模型,设计自适应伪谱遗传算法（APGA）,用以求解满足调整时间最优性的敏捷卫星任务规划问题。最后,通过仿真实验,验证了模型和算法能够有效地解决传统算法求解敏捷卫星任务规划问题时存在的求解效率低的缺陷。