【作者】 朱学彪；

【导师】 陈奎生；

【作者基本信息】 武汉科技大学 ， 机械设计及理论， 2013， 博士

【摘要】 液压伺服系统是目前工业、军事、航空航天等领域装备中的关键部件，其静、动态特性直接影响到设备的性能。随着对其动静态响应、输出功率密度和精度的要求不断提高，其结构动力学性能以及负载扰动、参数变化等不确定性因素的影响已成为不可忽略的问题，这些问题对液压伺服系统的结构设计和控制设计提出了巨大的挑战。本论文依托国家重大科研项目，面向重大工程的实际需求，从液压伺服系统的机理建模和鲁棒参数辨识、控制方法设计到原理样机实现三个方面展开研究，旨在通过鲁棒迭代控制来提高液压伺服系统的性能，使其满足制造和装配领域设备日益苛刻的性能需求。本文针对具有不确定性的轧机HAGC液压伺服系统，围绕其动力学建模与鲁棒控制中的关键问题，采用理论推导-仿真分析-实验测试相结合的方法，建立了轧机HAGC系统的参数化数学模型，研究了基于已知模型的鲁棒迭代学习控制器，并通过搭建相关测试实验平台进行了验证。在分析了轧机HAGC系统结构组成和体系特点的基础上，推导了轧机HAGC的机理模型，针对系统的不确定性，提出了一种基于Hardy空间的二段式非线性鲁棒辨识方法，得到一个参数待估的可行参数集合，建立其含参数不确定性的系统模型集，确保真实系统落在该模型集中。为了实现高精度高效率的运动，提出了一种基于斜坡正弦波函数的平滑轨迹规划方法，可以综合考虑系统的快速性和平稳性。针对液压伺服系统重复性工作空间，引入了反馈－前馈形式的迭代学习算法，设计了基于脉冲响应矩阵的迭代学习控制策略，使迭代学习控制器能够更好地反映系统动态特性，并且根据系统的误差输出来求取所需的控制量，刷新到迭代学习前馈指令中。综合考虑液压伺服系统模型的不确定性和系统稳定性指标、抗干扰指标，研究了基于定量反馈理论的液压伺服系统鲁棒控制器设计问题。讨论了适合液压伺服系统的迭代学习控制器(ILC）的设计方法，提出了一种基于H_∞法的鲁棒ILC设计方法，分析了该方法的可解性，推导出了误差收敛的充要条件。鲁棒ILC方法不仅将迭代学习控制器的综合问题转化为H_∞最(次）优控制器的综合问题，还可通过选择适当权函数，明确处理过程中不确定性因素。鲁棒ILC方法可通过μ综合方法来进行求解，使学习性能最大化。根据所提出的设计方法设计出了标称ILC和鲁棒ILC，在实验中分别执行了这两种ILC，根据实验结果对两种ILC进行了对比分析，实验结果验证了所提出设计方法的有效性。本文的研究成果成功地应用于轧机液压自动厚度控制(Hydraulic Automatic GaugeControl，HAGC)，完成了HAGC测试平台的搭建和控制系统软硬件设计，并在测试平台上成功进行了辨识和控制实验，为HAGC系统的结构设计和控制提供了分析依据和设计指导。更多还原

【Abstract】 Hydraulic servo systems are the important subsystems for the industrial, military andaerospace manufacturing. The static and dynamic characteristics of hydraulic servo systemsaffect the equipment performance directly. The structural dynamics, load disturbance andparameter changes of hydraulic servo systems cannot be ignored, with the increasingrequirements of quick response, output power density and motion precision. All of these posedtremendous challenges on the structure design and motion control of hydraulic servo system. Inorder to fulfill the urgent requirements of state key scientific research and engineering projects,the modeling and robust identification of hydraulic servo system, design of robust iterativelearning control and realization of principle prototype are studied in this dissertation. Throughthe characteristics analysis and controller synthesis, the performance of hydraulic servo system isimproved to meet the increasingly stringent performance requirements.In this thesis, the key issues of dynamic modeling and robust control for the hydraulic servosystem are studied. The parameterized mathematical model of hydraulic servo system isestablished by the combination method of theoretical analysis, simulation and experimentaltesting. The design of the controllers takes into account several aspects of the system’s dynamicsare completed. The proposed optimal approach is validated by the component level experimentalmeasurement. On the basis of analysis of the hydraulic servo system structures and systemcharacteristics，the mechanism model of the hydraulic servo system is derived. The two-stepnonlinear robust identification method based on Hardy space is proposed for the uncertainties ofthe system. The feasible parameter set of model parameters is obtained. The system model withparameter uncertainty set is established, to ensure that the real system falls on the model set. Inorder to achieve high precision and high efficiency motion, a smooth trajectory planning methodbased on ramp sinusoid functions, which can take the rapidity and smoothness of movement intoaccount.A feedback-feedforward form of iterative learning algorithms is introduced for the repetitivework space of hydraulic servo system. Iterative learning control strategy based on the impulseresponse matrix is designed to reflect the dynamic characteristics. According to the error of thesystem output to achieve the control signal, and refresh to iterative learning feed-forward order.Based on the quantitative feedback theory, the robust position controller for a hydraulic servosystem is studied, which considering model uncertain, stability and suppress disturbance ofmodel.Iterative learning control (ILC) design method for hydraulic servo system is discussed. Arobust ILC design method based H_∞method on is presented, and solvability of the method isanalyzed, the necessary and sufficient conditions of error convergence are derived. The robustILC method not only transformed iterative learning controller synthesis into most (sub) optimalcontroller design, but also deal with the uncertainties by selecting the appropriate weightfunction. The robust ILC method can be solved by the μ synthesis methods to maximize thelearning performance. The nominal ILC and the robustness ILC are design and performed in the experiment. The two ILC experimental results are compared, and the effectiveness of theproposed method is verified.The research results are applied to a hydraulic automatic gauge control (HAGC) systemsuccessfully. The hardware and software system of HAGC test platform are designed andimplemented. The identification experiments and control experiments are completed in the testplatform. The proposed methods can be used to guide the controller design for HAGC.更多还原