大型柔性航天器时变动力学参数在轨辨识方法

发布时间：2018-02-26 18:49

  本文关键词： 线性时变系统 模态参数辨识 状态空间模型 子空间辨识方法 大型柔性航天器　出处：《大连理工大学》2016年博士论文　论文类型：学位论文

【摘要】：大型柔性航天器结构通常无法在地面进行全尺寸动力学参数辨识实验,而当利用有限元方法对结构进行动力学分析时,由于模型简化、计算误差等影响,难以得到准确的模型参数。在这种条件下,采用在轨辨识方法,可以更为准确地获得航天器系统的动力学参数,从而反映航天器结构的真实运行情况。此外,航天器在轨运行时,大型附件运动、推进剂消耗等因素都可能使得结构的动力学参数发生持续变化。因此开展航天器时变动力学参数的在轨辨识方法研究,不仅可以对已建立的数值模型进行校验,还可以为航天器的控制系统设计与调整提供参考依据,具有重要的工程应用价值。本文考虑大型柔性航天器在轨运行时的动力学参数(模态参数和状态空间模型参数)时变情况,针对航天器周期变化动力学参数的辨识、时变模态参数和状态空间模型参数的递推辨识、以及航天器闭环系统的时变动力学参数辨识等内容开展了系统地研究。论文主要包括以下几部分内容：(1)研究航天器周期变化动力学参数的在轨辨识问题。考虑航天器大型柔性附件转动对系统动力学参数带来的周期性影响,利用周期子空间方法代替通常的重复实验方法,对柔性航天器周期时变的模态参数和相应的状态空间模型参数进行在轨辨识。仿真结果证明这种方法能有效辨识航天器的时变动力学参数。(2)研究航天器时变模态参数的递推在轨辨识问题,提出了一种提高TW-API递推子空间方法计算效率的改进方法。首先,针对基于奇异值分解(SVD)的辨识方法计算量较大的问题,采用基于信号子空间投影理论的投影近似子空间追踪(PAST)、逼近幂迭代(API)和截断窗逼近幂迭代(TW-API)等三种递推子空间方法辨识航天器时变模态参数,并对这三种方法的适用条件和计算效率进行了总结和比较。然后提出了一种改进的TW-API方法以提高在轨辨识的效率,该方法简化了数据处理中的矩阵递推过程,在保证计算精度的同时,显著地减少了在轨辨识过程的计算量,尤其当系统模型的阶次较高时,计算效率的优势更加明显。在数值仿真中,对改进TW-API方法和基于SVD的周期子空间方法的计算效率进行了比较,证明了递推子空间方法能够有效辨识航天器的时变模态参数,而且具有更高的计算效率。最后,针对上述几种递推子空间方法在低信噪比时存在的辨识结果误差较大的问题,利用小波降噪技术对信号进行去噪处理,获得了良好的辨识结果。(3)研究航天器时变状态空间模型参数的递推在轨辨识问题,提出了一种辨识状态空间模型参数中时变输入矩阵的新递推格式。不同于常用的重复实验辨识方法,新的递推格式基于信号子空间投影原理,通过重新建立输入-输出数据之间的关系,构建新的信号子空间矩阵以递推得到系统的时变输入矩阵,从而获得完整的系统时变状态空间模型参数。与现有的基于重复实验的辨识方法相比,新的递推格式不需要进行奇异值分解计算,有效减少了计算时间。(4)研究航天器闭环系统的时变动力学参数在轨辨识问题,提出一种辨识与校验时变输出反馈增益矩阵参数的新递推格式。新的递推格式通过构建系统的增广矩阵,利用最小二乘方法计算得到系统的时变输出反馈增益矩阵。这种递推格式避免了在直接使用最小二乘方法求逆时可能导致的矩阵参数辨识结果不唯一的问题,可以对设计的反馈增益矩阵参数进行检验。仿真结果表明提出的递推格式能够有效辨识航天器的时变输出反馈增益矩阵参数。此外,还对上述的周期／递推子空间方法在航天器闭环系统中的应用进行了研究,为闭环系统的时变动力学参数辨识问题提供重要的参考。
[Abstract]:Large flexible spacecraft structures usually cannot be on the ground for parameter identification of full size dynamics, and when using the finite element method for dynamic analysis of structures, the simplified model, calculation error, it is difficult to get the accurate model parameters. In this condition, the on orbit identification method can more accurately obtain the kinetic parameters of spacecraft the system, which reflects the real operation of the spacecraft structure. In addition, the spacecraft in orbit, large attachments, fuel consumption and other factors may make the structure of the dynamic parameters of persistent change. So to carry out the spacecraft in orbit identification method of time-varying parameters, can not only verify the numerical model has been established, also can provide a reference for the design and adjustment of spacecraft control system, has important engineering application value. This paper test The kinetic parameters considered for large flexible spacecraft on orbit (the modal parameters and the state space model parameters identification for time-varying), dynamics of spacecraft cycle, recursive identification of time-varying modal parameters and the state space model parameters, and the spacecraft of the closed-loop system and dynamic parameter identification of time-varying system etc. to study. This paper includes the following parts: (1) the change of spacecraft Cycle Kinetic Parameters on orbit identification problem. Spacecraft large flexible rotation periodic impact on system dynamics parameters into consideration, using the periodic subspace method instead of the usual method of repeated experiments, on Orbit Identification of modal parameters of flexible spacecraft and the corresponding periodic time-varying the state space model parameters. The simulation results show that this method can effectively identify the time-varying dynamics of spacecraft. Number. (2) recursive problems in orbit identification of modal parameters of spacecraft when the method is proposed to improve computational efficiency of recursive subspace method to improve the TW-API. For the first, based on the singular value decomposition (SVD) involves a large amount of calculation of the identification method, using projection signal subspace projection based on the theory of approximate subspace space tracking (PAST), the power iteration approximation (API) and truncated window approximation power iteration (TW-API) and other three kinds of modal parameters recursive subspace identification method of spacecraft, and the suitable conditions and the calculation efficiency of the three methods were summarized and compared. Then we propose an improved TW-API method to improve efficiency in orbit identification, this method simplifies the matrix data processing in the recursive process, the accuracy of the guarantee, significantly reducing the amount of calculation in orbit identification process, especially when the high order system model, calculation The efficiency is more obvious advantages. In numerical simulation, the improved TW-API method and the calculation efficiency of periodic subspace methods based on SVD were compared. The modal parameters proved that recursive subspace method can effectively identify the spacecraft, but also has higher computational efficiency. Finally, according to the results of the several recursive subspace identification method in low signal-to-noise ratio error problem of signal denoising by wavelet de-noising technology, obtained good identification results. (3) recursive problems in orbit identification of state space model parameters of the spacecraft, a state space model parameter identification of time-varying input matrix the new recursive scheme. Repeated experimental identification method is different from the commonly used recursive scheme, signal subspace projection based on the new principle, through the re establishment of the relationship between input and output data, Construction of the new signal subspace matrix to achieve recursive system time-varying input matrix, so as to obtain the complete system time-varying state space model parameters. With the existing identification method based on repeated experiments compared with recursive scheme does not require singular value decomposition calculation, effectively reduce the calculation time. (4) study the spacecraft of the closed-loop system time-varying dynamics parameters on orbit identification problems, put forward a method to identify and verify the time-varying output feedback gain matrix parameters of the new recursive scheme. Recursive scheme through augmented matrix system of the new construction, obtained by least square method of time-varying output feedback gain matrix. The recursive scheme to avoid the possibility of lead in the direct use of least squares method to solve the inverse of the matrix parameter identification results not only problem, can test the feedback gain matrix design parameters. The results of the simulation table The proposed recursive scheme can effectively identify the spacecraft time-varying output feedback gain matrix parameters. In addition, the cycle / recursive subspace method of the application in spacecraft closed-loop system are discussed, the closed-loop system with time-varying dynamic parameter identification problem provides an important reference.

【学位授予单位】：大连理工大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：V414
，

本文编号：1539255