【作者】 曹璐；

【导师】 陈小前；

【作者基本信息】 国防科学技术大学 ， 航空宇航科学与技术， 2010， 硕士

【摘要】 现代小卫星以其高性能、低风险、短周期、低投入等特点,成为空间技术发展的一个重要方向。但由于受到成本、尺寸、质量、功耗等条件的限制,使得小卫星常常采用低精度、低功耗敏感器进行姿态确定及控制,致使姿态确定及控制水平普遍较低,严重影响了小卫星的总体性能。为了更好的提高小卫星性能,满足未来任务的需求,研究基于低精度敏感器实现高精度姿态确定及控制的技术,是有着重要意义的。论文正是以此为背景,研究解决了以下几个关键问题,主要内容如下:研究了多种改进的EKF姿态确定方法。在对传统的EKF方法进行了深入研究的基础上,通过对观测矢量的进一步推导,获得了基于角速度的观测方程,丰富了观测模型,提高了观测数据的利用率,并在此基础上对姿态确定流程进行了设计,改进了传统EKF算法,并通过仿真,获得了较高的姿态确定精度。研究了联邦滤波算法和预测滤波算法并将其应用于小卫星姿态确定中。通过对联邦滤波算法的研究,提出了基于新息观测的信息分配因子方法,该方法通过将观测新息引入到信息分配因子中去,提高了信息分配因子的准确性和可靠性,进而提高了改进联邦滤波算法的姿态确定精度。通过对预测滤波算法的研究,提出了将模型误差及角速度误差同时进行解耦估计的预测滤波策略,提高了模型的准确性,加速了角速度的收敛速度,进而增强了改进预测滤波算法处理非线性、非高斯白噪声的能力,通过仿真验证了两种方法均具有较高的滤波精度。研究了粒子滤波加速方法并将其应用于小卫星姿态确定中。通过对粒子滤波理论的学习、研究,大胆引入极限近似的思想,以此证明了采用少量粒子进行状态估计的可行性,并将“局部设计”的思想引入贝叶斯滤波及序贯重要性采样方法中,通过设计混合重采样方法进一步完善了整个粒子滤波加速方法,并在此基础上,对其进行了未知参数估计的增广研究。最后以小卫星姿态确定为例,提出了运用粒子滤波加速方法进行状态估计的一般流程,并取得了较前文方法更高的滤波精度。研究了基于滑模变结构控制的小卫星姿态控制方法。研究了基于I/O线性化的滑模变结构模糊自适应控制方法,取得了较好的控制效果,在此基础上,通过对非奇异Terminal滑模变结构的设计,提出了非奇异Terminal滑模变结构模糊自适应控制,验证了该方法可以在有限时间内实现姿态控制的快速收敛,通过进一步对该思想的完善,提出了全局快速积分Terminal滑模自适应控制,不仅使小卫星的姿态在有限时间内实现快速收敛,同时增强了其抗干扰能力。最后,通过仿真验证了控制方法的可靠性和有效性。论文研究的非线性滤波方法及滑模变结构控制理论,具有较强的非线性处理能力,可以实现低精度敏感器的高精度姿态确定及控制,有助于提高小卫星的总体性能,论文的研究工作可以为这一领域的技术研究和发展提供方案参考和技术支持。更多还原

【Abstract】 With fast development of space technology, modern small satellite plays a more and more important role due to its advantages of high performance, low-risk, short periods and low-cost. However, sensors with low precision and power consumption are usually adopted for small satellite because of the limitations on cost, size, mass and power, which decreses the level of attitude determination and control system (ADCS) and the whole performance. To solve this problem, methods for high precision attitude determination and control with low precision sensors are presented in this dissertation.The key technologies presented here are listed as following:Several EKF based attitude determination methods are studied. Comprehensive research on traditional EKF is carried out, and an observation equation is derived from observation vectors, which can enrich the traditional observation equation and enhance the utilization efficiency of observation data. Then the flow of attitude determination is designed and several improved EKF algorithms are proposed, all of which can achieve better accuracy.The application of Federal Filter and Predictive Filter on snall satellite attitude determination is investigated. To enhance the accuracy and reliability of information distribution gene in Federated Filter, a new method of information distribution gene is proposed based on better utilization of new observation information; so as to further improve the attitude determination precision. To improve the mode accuracy and the convergence efficiency of angle velocity in Predictive Filter, a new approach to estimate the mode error and angle velocity error concurrently by decoupling is proposed, which can better handle nonlinear and non-Gaussian white noises. The high performance of filter precision of these two new proposed methods are verified by simulation results.The accelerating Particle Filter method and its application in small satellite attitude determination are discussed. The idea of limit approximation is introduced into to the Particle Filter theory, which proves the feasibility of estimating state variables with fewer amounts of particles. To enhance the accelerating Particle Filter method, a new mix-resampling method is put forward by introducing“local-design”into the theory of Bayes Filter and Sequential Importance Sampling, based on which extension Particle Filter to estimate unknown parameters is further studied. At last, a general flow of parameter estimation by accelerating Particle Filter is developed, and tested in the example of small satellite attitude determination, which verifies its high performance of filter precision and even better compared with the aforementioned approaches.The small satellite attitude control based on sliding mode control is studied. The I/O linearization based adaptive fuzzy sliding mode control is firstly investigated, which achieves good control performance. Based on this method, a non-singular terminal adaptive fuzzy sliding mode control approach is proposed, and its capability of fast convergence in attitude control within limited time is proved. To perfect this research, a new control policy of global fast integral terminal sliding mode control is put forward, which can better improve fast convergence speed and anti-jamming effect. The three foregoing control policies are combined with accelerating Particle Filter and tested by simulations, which have verified their reliability and validity.The nonlinear filter method and variable structure control theory proposed in this dissertation have the powerful capacities to deal with nonlinearity and benefit to implement high-precision attitude determination and control with low-precision sensors in order to improve the system performance for small satellite. The results in this dissertation can provide reference and technique support for future research in this field.更多还原