【作者】 王鹏；

【导师】 张迎春；

【作者基本信息】 哈尔滨工业大学 ， 控制科学与工程， 2008， 博士

【摘要】 随着中国神舟系列飞船的成功发射、探月计划和深空探测计划的实施,航天器自主生存能力成为新世纪备受关注的关键技术之一。航天器自主导航及姿态确定技术是卫星自主性的一个重要方面,是当今航天器控制技术的发展趋势,它在减轻地面测控系统负担、降低航天器运行费用、提高航天器生存能力以及扩展卫星的应用潜力等方面都具有重要意义。因此利用多种星载敏感器采用信息融合的方法提高卫星自主导航及姿态确定的精度和可靠性是目前航天器设计的关键内容与热点问题。本论文结合国家安全重大基础研究项目(973)“微型航天器新概念、新机理研究(51312)”这一任务,研究了基于各种星载敏感器相结合的卫星自主导航及姿态确定方法,具体工作如下:与传统方法利用磁强计测量地磁场信息仅单一确定行航天器导航参数或姿态信息相比,提出一种将陀螺与磁强计相结合利用地磁场测量信息同时确定低地球轨道卫星导航参数与姿态信息的方法。首先详细分析了地磁场数学模型及磁强计观测模型;然后推导了轨道六要素轨道动力学模型及姿态四元数运动学模型,建立组合系统状态方程;其次提出将磁强计测量值与国际地磁参考场(IGRF)模型估计值作差,建立组合系统观测方程,通过对两者观测差值的微分分析得到:只用一个观测表达式即能够同时包含航天器的导航参数及姿态信息,最后设计了先进的连续-离散扩展卡尔曼滤波器对组合系统进行了数值仿真,并对仿真结果及系统性能进行分析和讨论。针对目前利用单一天体敏感器自主导航的缺点和不足,提出三种基于信息融合的自主天文导航方法:(1)根据星敏感器与地球敏感器相结合敏感地平方式的不同可以分为直接敏感地平与间接敏感地平。当间接敏感地平观测不到折射星时,引入由地球敏感器测量得到的卫星距离及方向矢量作为新的观测信息。(2)太阳敏感器与磁强计相结合,当卫星运行在不同区域时,根据太阳敏感器的工作原理,可以分为太阳光照区和太阳阴影区两种工作模式。当卫星运行在太阳阴影区时,引入国际地磁参考场(IGRF)模型,以磁强计测量值与IGRF模型夹角余弦作为新的观测信息。(3)针对上述所提出两种基于信息融合的自主导航方法的缺点和不足,提出将上述卫星上常用的天体敏感器相结合进行自主导航。根据上述三种自主导航系统各自的特点,分别设计了基于信息融合的自适应EKF算法对上述系统进行数值仿真,并对仿真结果及系统性能进行分析和讨论。捷联惯性导航系统(SINS)是一种常用的自主导航系统,具有自主性、隐蔽性、宽频带和信息全面等优点,但是其惯性测量元件误差随着时间累积、难以长时间连续工作的缺点和不足同样是非常明显的。星敏感器是目前应用最广泛的星载姿态敏感器,具有指向精度高、无姿态积累误差等特性。因此提出将捷联惯性导航系统与星敏感器相结合,目前最常用的SINS系统误差数学模型是在地球固联坐标系或地理坐标系中的,而星敏感器观测恒星是在地心惯性坐标系中,所以推导了在地心惯性坐标系下的SINS误差数学模型及其传播特性。选取40维状态变量建立组合导航系统状态方程;由SINS和星敏感器输出的姿态四元数得到姿态四元数误差,建立组合导航系统观测方程,最后采用先进的连续-离散扩展卡尔曼滤波器进行数值仿真,并对仿真结果进行分析。系统状态的可观测性和可观测度是检验所设计滤波器收敛精度和速度的重要指标。针对前述所提出的典型自主导航系统,提出三种分析系统可观测性和可观测度的方法:(1)根据系统可观测性定义及分析方法,对直接敏感地平天文导航系统可观测性及影响其的主要因素进行了数值分析;(2)对分段线性定常系统(PWCS)可观测性进行了分析;(3)对PWCS可观测性矩阵进行奇异值分解,分析了系统状态的可观测度。对SINS/CNS组合导航系统进行了基于奇异值分解的可观测性和可观测度分析,对降维前后组合导航系统进行数值仿真,并对仿真结果和系统性能进行了分析和讨论。更多还原

【Abstract】 Along with the Shenzhou series spacecraft, the lunar exploration project and the deep space exploration Project, the self-surviving performance of the spacecraft is regarded as one of the key technologies in 21st century. The autonomous navigaition and attitude determination technology is not only important aspect of spacecraft autonomy but also development trend of spacecraft control technology. It is important meaning for the modern satellite to reduce burden of ground-assisted system, decrease the cost of performance missions, greatly strengthen the probe’s survival ability and extend the potential space application of the spacecraft. One of the new and widely researched problems in the subject of spacecraft design is to autonomously determinate orbit parameters by using measurement information of multi-spaceborne sensors and to improve system precision and reliability based on using information fusion method. With the support of the Nation Security Magnitude Foundation Research Foundation–Research on the new concept and new mechanism of minisize spacecraft, this dissertation deals with automous navigation and attitude determination based on multi-spaceborne sensors. The main contributions are as follows:The tradition methods using measurement of geomagnetic field based on magnetometer only determinate navigation or attitude parameter in singles. The method of determinating navigation and attitude parameter of LEO satellite simultaneitily using geomagnetic field information is presented based on integrated system of magnetometer and gyro. Firstly, the geomagnetic mathematic model and magnetometer measurement model are analyzed in detail. Secondly, the state equation of integrated system is established on deducing the model of the oribt dynamic model based on the orbit six elements and attitude kinematic model based on attitude quaternion. Thirdly, the observed equation of integrated system is established by using the difference of the measurement value of magnetometer and the estimated value of IGRF model. The observed equation includes spacecraft navigation and attitude parameter by analyzing the difference value synchronously. Compared with traditional method, the new method can estimate the spacecraft navigation and attitude parameter synchronously. Finally, the continuous-discrete extend kalman filter(CDEKF) algorithm is designed based on integrated system. The simulation result and system performance are analyzed and discussed after numerical simulation.Contraposing disadvantage and deficiency only using single spaceborne celestial sensor, the three methods of autonomous celestial navigation are presented based on information fusion. Firstly, the sensing earth modes include directly sensing horizon and indirect sensing horizon. When the refraction stars can’t be observed, the distance and direction vector from satellite to earth is imported as the new observation information. Secondly, autonomous navigation method is presented based on sun sensor and magnetometer. According to the operational principle of sun sensor, the operational mode can be divided into two parts: sun shining area form and sun shadow area form. The estimated value of IGRF model is imported as the new observation information. The new observation value is got by the angle cosine based on the measurement value of magnetometer and the estimated value of IGRF model. Finally, Contraposing disadvantage and deficiency of the above-mentioned methods, a new autonomous navigation method is presented based on multi-celestial sensors. The Federal Extended Kalman Filter algorithm based on information fusion method is designed by analyzing the characteristic of aforementioned antonomous navigation methohds. The simulation result and system performance is analyzed and discussed after numerical simulation.The strapdown inertial navigation system(SINS) is a autonomous navigation system including the advantage of independence, concealment,wide frequency band and all-around information. As the drift errors of the inertial measurement component increase with time, the disadvantage and deficiency of SINS is obvious, which is unsuitable for long-time operating continuously. As a kind of used widely spaceborne sensor, the star sensor gains popularity for its high accuracy, without accumulative and intelligence. A new integrated navigation mode of SINS and star sensor is presented to correct the navigation parameter. The SINS error mathematic model is in the earth fixed coordinate or the geography coordinate usually, then the star sensor observes navigation stars in earth-centered inertia coordinate. The error mathematic model and diffusion characteristic of SINS are deduced in earth-centered inertia coordinate. The state equtation of integrated navigation system is established by 40 dimension state variables. The observation equtation of integrated navigation system is established by attitude quaternion error defined by attitude quaternion of SINS and star sensor. The integrated navigation system of SINS/CNS simulates by aforementioned CDEKF and the simulation results are analyzed.The observability and observable degree of the system state variables are the key indicators to check the convergence accuracy and velocity of designed filter. The three analysis methods of the system observability and observable degrees are presented contraposing aforementioned typical autonomous navigation systems. Firstly, the observability and influence factor of the directly sensing horizon celestial navigation system is analyzed based on the observability and observability degree. Secondly, the observability of piece-wise constant system(PWCS) is analyzed. Finally, the observability and observable degree of integrated navigation system of SINS/CNS is analyzed by the singular value decomposition and descend the integrated navigation system dimensions. The simulation result and system performance is analyzed and compared with undescended system and descended system after numerical simulation.更多还原