QQ客服 官方微博 反馈留言

题名	空间遥操作系统的预测控制方法研究
作者	杨艳华
学位类别	博士
答辩日期	2013-11-20
授予单位	中国科学院沈阳自动化研究所
导师	李洪谊
关键词	空间遥操作 时变时延 预测控制 混合H2/H∞ 自适应控制 LMI
其他题名	Model Predictive Control for Space Teleoperation System
学位专业	模式识别与智能系统
中文摘要	随着我国航空航天研究的发展和探月工程的实施，我们将面临许多更加复杂的空间任务，比如星球表面探索、空间装配等。如果利用空间机器人代替宇航员完成这些复杂的空间任务，不但可以减小宇航员的危险性，还可以降低成本。然而，受现有机器人技术水平的限制，目前还很难研制出能够在复杂多变、非结构化的空间环境中完全自主运行的空间机器人。应运而生遥操作技术使得人们能够操纵远端的机器人完成相应的任务，因此，目前遥操作是应用空间机器人的主要形式，空间遥操作也因此成为当前研究课题的热点之一。但是，在空间遥操作系统中，信号的长距离传输和带宽的限制往往会引起较大的时变时延。这严重影响了遥操作系统的操作性能，甚至使得系统不稳定。此外，环境模型的不确定性也使得空间遥操作系统的控制变得更为复杂。本文在分析了现有遥操作控制技术和理论的基础上，针对空间遥操作系统的大时变时延和模型参数不确定问题展开了深入研究。主要研究内容包括对时延的分析和预测，以及在此基础上提出的空间遥操作系统的控制方法。首先，分析了Internet 网络时延的统计特性，并且提出了一种基于稀疏矩阵的核非线性回归方法来预测网络时延，从而提高了网络时延的预测精度；并且将该方法用于天地时延中地面网络时延的分析和预测，预测结果表明了该方法的有效性。其次，针对空间遥操作系统的时变大时延问题，提出了一种基于在线时延预测的广义预测控制(Generalized Predictive Control, GPC)方法。在主端设计时延预测器和GPC，在每个采样时刻，根据预测的时延值动态调整GPC的预测时域，从而保证系统在大时变时延下的稳定性。此外，根据空间遥操作系统仿真实验平台主从端模型未知的情况，在上述方法的基础上，提出了一种无需知道主从端模型的基于时延预测的模型预测控制(Model Predictive Control, MPC)算法，并且在遥操作实验平台上进行了实验验证。然后，为了提高空间遥操作系统的鲁棒性，提出了一种基于线性矩阵不等式(Linear matrix inequality, LMI )的鲁棒H∞控制方法。通过定义Lyapunov函数，分别分析了环境模型参数确定和不确定情况下的闭环系统稳定性。而且，针对非对称的网络时延和主从端机械臂模型是不确定、非线性的空间遥操系统提出了一种自适应H∞控制方法。该方法将非线性的主从端模型转化为不依赖于主从端动力学参数的线性模型，而且将遥操作系统的稳定性和透明性问题转化为具有多输入时延的线性系统的H∞状态反馈控制问题。通过定义Lyapunov–Krasovskii函数和自由权矩阵，给出了系统渐近稳定且满足给定的位置和力跟踪性能的充分条件。仿真实验验证了该方法的有效性。最后，为了提高MPC控制方法的鲁棒性以及解决现有用于遥操作系统的H∞控制方法需要事先假设时延的上下界可能引起的保守性和不稳定性问题，将MPC和H∞控制方法相结合，针对具有大时变时延和控制输入受限的力反馈空间遥操作系统，提出了一种基于混合H2/H∞空间遥操作系统的MPC方法。该方法无需事先假设遥操作系统回路时延的上下界，而且既能够补偿大时变时延，满足控制输入约束条件，又能够满足一定的跟踪性能指标。将该方法分别同MPC和H∞控制方法进行比较，比较结果表明该方法的优越性。本文的工作丰富了空间遥操作系统的研究内容，针对空间遥操作系统目前存在的稳定性和鲁棒性问题提出了多种控制方法，为后续的研究提供了理论基础和技术保障。
索取号	TP273/Y27/2013
英文摘要	With the development of chinese aerospace research and the implementation of the lunar exploration project, we will be faced with many more complex space missions, such as the planet's surface exploration, space assembly, and so on. If these complex space missions are completed by space robots instead of astronauts, it will not only reduce the risk of astronauts, but also decrease costs. However, due to the limitations of existing robotics level, it is still difficult to be able to develop a completely autonomous space robot under the complex, unstructured space environment. Emerged teleoperation technology allows people to accomplish tasks by operating the remote robot. Therefore, at present, space teleoperation is the main form of the application of space robots and space teleoperation has thus become one of the hot topics of current research. However, in the space teleoperation system, long-distance transmission of signals and limited bandwidth often lead to large time-varying delay, which seriously decreases the operational performances, and even makes the system unstable. In addition, uncertain environment models also make the control of space teleoperation systems more complex.After analyzing the existing teleoperation control technologies and theories, this paper studies the issues about large time-varying delay and model parameter uncertainty in space teleoperation systems. The main contents include time delay analysis and forecasting, and control methods for space teleoperation systems.First, after analyzing the statistical properties of Internet delay, we propose a sparse matrix based on kernel non-linear regression (SMKR) scheme to predict Internet delay and it effectively improves the prediction accuracy. In addition, it is used to analyze and predict the ground delay of space-ground time delay. Prediction results demonstrate the validity of this method.Second, for the problem of large time-varying delay in space teleoperation system, we propose a generalized predictive control (GPC) method based on on-line time delay prediction. On the master side, a round trip time delay predictor and a GPC are designed. At each sample time, the prediction horizon of GPC is dynamically adjusted based on the predicted delay, which ensures teleoperation system stability under large time-varying delays. In addition, because the master and slave manipulator dynamics models of a space teleoperation system experimental platform are unknown, according to the above method, a model predictive control (MPC) algorithm based on time delay prediction is proposed, which not need to know the master and slave manipulator model. Some experiments are done in the platform for verifying the validity of this method. Then, in order to improve space teleoperation system robustness, a robust H∞ control method is proposed based on linear matrix inequalities (LMI). By defining Lyapunov function, the closed-loop system stability is analyzed respectively under the cases that the model parameters of environment are certain and uncertain. Moreover, an adaptive H∞ control method is proposed for the space teleoperation systems, in which network delay is non-symmetric and time-varying and the master and slave manipulator models are uncertain and nonlinear. In this method, the master and slave nonlinear models are respectively transformed into linear models, which do not depend on the kinetic parameters of the manipulators. Moreover, the problem of space teleoperation system stability and transparency is converted into an H∞ state feedback control problem for a linear system with multi-input delay. By defining Lyapunov-Krasovskii function and free weight matrixes, sufficient conditions are given such that the system is asymptotically stable and satisfies the given position and force tracking performance. Simulation results demonstrate the effectiveness of this method.Finally, existing H∞ control methods for teleoperation systems require prior assumptions the bounds of time delays, which may cause conservative and instability problem. In order to solve the problems and improve the robustness of the MPC control method, a MPC method based on hybrid H2/H∞ is proposed for space teleoperation systems that are with unknown, large time-varying delays and input constraints. This method does not require assuming the delay bounds. Moreover, H∞ control approach has been incorporated into the MPC, which compensates large time-varying delay, handles input constraints, and satisfies tracking performances. The methods are respectively compared with MPC and H∞ control methods. The simulation results show that the method has better performance than others.This work, which enriches the research of space teleoperation system, proposes a variety of control methods for space teleoperation system, which currently has stability and robustness issues. It provides a theoretical foundation and technical support for the following tasks.
语种	中文
产权排序	1
页码	104页
分类号	TP273
内容类型	学位论文
源URL	[http://ir.sia.ac.cn/handle/173321/14839]
专题	沈阳自动化研究所_机器人学研究室
推荐引用方式 GB/T 7714	杨艳华. 空间遥操作系统的预测控制方法研究[D]. 中国科学院沈阳自动化研究所. 2013.

个性服务

查看访问统计

相关权益政策

暂无数据

收藏/分享

除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。