遥操作机器人系统的智能控制研究

【作者】 陈启宏；

【导师】 费树岷；

【作者基本信息】 东南大学 ， 自动控制理论及应用， 2003， 博士

【摘要】 基于预测控制与Lyapunov泛函方法，用神经网络、线性矩阵不等式(LMI)等工具、以稳定性和透明性为主要目标，讨论遥操作机器人系统的各种反馈控制器的设计方法。全文主要由以下几部分组成：第一部分 首先介绍遥操作机器人系统的发展历程、研究概况、在遥操作机器人系统的研究中所涉及到的一些主要方法与工具。然后提出本文研究的一些主要问题和本文研究所需的一些准备知识。第二部分针对基于Internet的遥操作机器人系统的时变通讯时延和未知、时变、甚至非线性环境模型，利用前向神经网络建立主机械手、从机械手和环境的模型。通过神经网络模型预测主机械于未来位置、速度和从机械手受力，使得从机械手能实时跟踪主手位置和速度，提高系统的操作性能，并结合无源控制算法进行控制切换，保证系统的稳定性。最后对Internet的通信协议、时延和数据丢失进行了分析和处理。第三部分针对遥操作机器人系统传输通道中的通讯时延和系统模型的不确定性，造成系统不稳定和操作性能降低等问题，提出用带记忆的位置和速度反馈控制方法，并用线性矩阵不等式对系统的反馈参数进行设计，使系统鲁棒渐近稳定，主、从机械手静态跟踪误差较小，而且使系统具有良好的透明性。考虑到这种方法的可解性并不是太强，在此基础上，本文还提出用带记忆的力反馈控制方法，使得系统稳定、透明性良好，而且系统参数易解。第四部分提出用时间前向观测器预测从手状态，用力、位置和速度反馈消除或减小时延对系统影响的方法，并对反馈参数的设计进行分析。时延时变时，我们将时延的变化率建模为系统不确定参数或扰动，分四种情况(时延时不变、环境模型已知；时延时不变、环境模型未知;时延时变、环境模型已知：时延时变、环境模型未知)讨论了系统的稳定性，而且使系统具有良好的透明性。本文对主要的设计方案进行了仿真研究。仿真结果表明，本文所给出的控制器设计方案均可以获得良好的控制效果。更多还原

【Abstract】 Based on predictive control and Lyapunov function, this dissertation devotes on the design of feedback controller for teleoperation robot systems by using neural network models and linear matrix inequalities to guarantee stability and transparency of the systems. The main work of this dissertation consists of four parts as follows:In part I , the development about teleoperation robot systems and general situation about their research are discussed first, and then the main methods and some tools, which will be used in this dissertation, are stated. Furthermore, the major problems, which are studied in this dissertation, are introduced. Finally, some necessary preliminary results in this dissertation are given.In part II ,in order to eliminate or reduce the effect of time-varying time delays, data loss and unknown, time-varying, even non-linear environment to an Internet-based teleoperation robot system, back propogation neural network is used to build the model of slave and environment. The master’s position, velocity and force exerted on the slave are predicted through the neural network models such that the slave can track the master real time. Moreover, Controllers are switched between the predictive control and passive control to make sure that the whole system is stable and transparent. At last, we analysis communication protocols of Internet and treat with time delays and data loss.In part III, memory position and velocity feedback is proposed to guarantee stability and transparency of a teleoperation robot system with time delay and uncertain parameters. Linear matrix inequality is used to design feedback parameters of the system. In this way, static track error between the master and the slave is little, however, sometimes the feedback parameters doesn’t exist. In order to solve this problem, we propose memory force, position and velocity feedback control method. The system can achieve good performance using this method and feedback parameters always exist.In part IV, time forward observer is used to predict states of the slave. Moreover, force, position and velocity feedback is proposed to design the system. Feedback parameters are solved through linear matrix inequalities. When time delay is time varying, the varying rate of the time delay is modeled as uncertain parameters or disturbance. Stability and transparency are discussed in four cases (time varying delay and known environment; time varying delay and unknown environment; time invariable delay and known environment; time invariable delay and unknown environment).In this dissertation, simulations are made for major design schemes. Simulation results show the effectiveness of the proposed approaches.更多还原