【作者】 闫丹丹；

【导师】 李元春；

【作者基本信息】 吉林大学 ， 控制理论与控制工程， 2004， 硕士

【摘要】 本文受吉林省杰出青年基金项目资助，对复杂系统的自适应容错控制问题进行了深入的研究。自动化系统日益大型化和复杂化，相应地对生产的可靠性要求也就越来越高，容错控制就是由此发展起来的一种提高控制系统可靠性的技术。而自适应方法能够有效的解决系统的不确定性和鲁棒性问题，所以在容错控制领域中，自适应控制方法有很大的发展前景。尤其在许多安全性比较重要的系统，例如飞行控制系统中，发生传感器或执行器故障时，我们无法获知有关故障的先验知识，采用自适应控制方法往往能获得令人满意的控制效果，即使在故障未知的情况下。1991年F.Ahmed-Zaid等人将自适应方法应用到F-16飞机的故障调节上。由于自适应方法本身会引起系统的非线性，其故障瞬间的过渡过程特性并不能令人满意。本文首先综述了各种容错控制方法，以及目前自适应方法在容错控制领域的应用，然后对系统的故障模型进行分析，建立执行器故障模型。对一类典型的多输入单输出复杂系统，设计基于Lyapunov方法的模型参考自适应执行器故障补偿控制器，理论上证明了采用所设计的自适应控制律能够实现系统的全局稳定性。本设计的控制目标就是设计一个控制器，即使在执行器发生故障时，仍然可以通过调节剩余的执行器来获得期望的性能指标。为了设计自适应控制策略，我们有必要了解当系统的参数和故障已知时，期望的控制器的结构的情况。研究故障发生前后的过程模型匹配条件，有助于任意执行器故障发生时控制器参数的自适应调节。我<WP=86>们分别针对参考模型的状态和输出设计相应的自适应控制律，通过不断的自适应调节状态反馈控制的各个参数，最终使得状态和输出都能够跟踪给定的参考信号。由于飞机系统是安全性要求比较高的典型系统，因此本文对一种线性化后的飞机系统模型进行了仿真研究，结果证明了设计方法的有效性。考虑到实际应用中，系统的内部状态并不都是可测量的，本文继续针对一类多输入单输出的复杂系统设计采用输出反馈的自适应控制策略。仍然对同样的执行器故障进行设计，仿真结果证明了所设计的方法能够实现系统的全局稳定性。但是由于自适应方法本身会引起系统的时变非线性，因此不能够保证系统的过渡过程特性，尤其是在未知执行器故障发生的瞬间，虽然系统最终保证了收敛性和稳定性，但过渡过程特性不能够令人满意。因此本文就改善系统故障瞬间的过渡过程特性，给出了一种带有固定补偿器的改进自适应执行器故障补偿控制方法，通过设计固定补偿器，组成一个附加的反馈环，来抵消由于参数不确定性和参数辨识所引起的误差，并对设计的自适应方法进行了性能分析，根据均方差和范数理论对该方法进行了证明。最后，对比没有固定补偿器的自适应方法和所设计的改进的带有固定补偿器的自适应方法，将其应用到线性化后的飞机系统模型，验证该改进方法针对方向舵执行机构故障能够有效的改善系统的性能指标，尤其是故障瞬间的过渡过程特性。　　综上，本文的主要工作内容如下：1．综述了容错控制发展的概况和现状，阐述了课题的背景及实际应用意义，并对自适应容错控制方法的分类及应用进行了深入的研究和详细的总结。2．分析控制系统各种故障模型的数学表示，并对于本文所要研究的多输入单输出系统进行执行器故障模型的建立。其中执行器的故障形式包括恒值(卡死)故障、可参数化的时变故障和不可参数化的时变故障。3．针对所建立的执行器故障模型设计状态反馈自适应故障补偿控制器。首先对自适应方法的模型匹配问题进行分析，给出前提条件；然后依据参考模型，利用状态反馈分别设计满足要求性能指标的状态跟踪和输出跟踪的自适应执行器故障补偿控制器；最后采用多输入单输出的飞机系统<WP=87>模型进行了仿真研究，仿真结果证明了所设计方法的有效性，所有的闭环信号都是有界的，系统的状态和输出最终都能跟踪给定的参考状态和输出。4．针对所建立的执行器故障模型设计输出反馈自适应故障补偿控制器。本章首先根据所建立的故障模式设计输出反馈的自适应故障补偿控制器，分析其结果可知所有的闭环信号都是有界的，系统输出最终都能跟踪给定的参考输出。但是，故障瞬间的过渡过程特性却并不能够令人满意，这是由于自适应方法会引起系统的时变非线性，不能够保证系统的过渡过程特性。因此本章在原有控制器的基础上设计了一种改进的自适应故障补偿控制器，能够改善故障瞬间系统的过渡过程特性。该控制器包含一个由固定补偿器组成的附加反馈环，能够抵消由于自适应方法中参数不确定性和参数辨识所引起的误差。最后进行了仿真研究，通过比较仿真结果，可以验证系统的过渡过程特性有了明显的改善。5．系统地对本文进行了总结，并指出进一步的研究重点。更多还原

【Abstract】 Based on the fact that automatic systems are becoming more and more large and complex and the control accuracy for the practical engineering systems is advanced as high as possible, fault tolerant control was a reliable technology and put forward to improve control performance. Adaptive control as an effective method can solve the problem of the systems with uncertainty and robustness, so this control method has a promising foreground in fault tolerant control fields, especially for some systems focusing on their safety such as airplane system. When something has been wrong with the actuator or sensor of the system, where nothing about the failure is known previously, adaptive control scheme is applied to the system and always can get satisfied performance, even under the condition that failure pattern is unknown. In 1991, adaptive method was applied to F-16 airplane for its failure accommodation by F.Ahmed-Zaid. But unfortunately, the major difficulty with adaptive methods is that the transient performance could not be assured due to the property of the nonlinear time-varying systems that they result in. First, in this dissertation all sorts of fault tolerant control methods and the present situation and development of adaptive schemes in fault tolerant control fields are summarized. Then failure models of the system are analyzed and presented.Model Reference Adaptive Controller (MRAC) based on the Lyapunov method as actuator failure compensator is designed for a class of <WP=89>continuous-time, multi-input and single-output linear time-invariant complex plants. Global stability is proved in theory for the design of the adaptive control law. The control task is to design a controller and accommodate the rest actuators in order to attain the desired performance under the condition that failure parameters are unknown. The parameters of the system and failure pattern need to be analyzed and something of the desired controller should be known in order to design the adaptive control scheme. The plant model match parameters are discussed before and after the failure happens, so the designed controller can adaptively regulate arbitrary actuator failure. According to the states and output of the reference model respectively, adaptive control schemes are presented. Adaptively parameters of the state feedback controller are regulated continuously, so ultimately the states and output both could follow after the given reference signals. Due to airplane system as a typical one, which focused much on its safety, simulation results for a linearized airplane model with actuator failure are shown to verify the efficiency of the designed adaptive methods.Considering the practical situation that the internal states are not always measurable, an adaptive control scheme based on output feedback is given for a class of continuous-time, multi-input and single-output linear time-invariant complex plants. According to the same actuator failure, simulation results for a linearized airplane model with actuator failure are shown to realize the global stability. Unfortunately, the major difficulty with MRAC method is that the transient performance could not be satisfied due to the property of the nonlinear time-varying systems that result in notwithstanding ultimately the convergence and stability of the system are assured. In this dissertation we will seek ways of improving performance of the adaptive failure compensation controller. A modified adaptive scheme including an additive feedback loop through fixed compensator, which is introduced to counteract the error caused by parameter uncertainty and the inaccuracy of parameter identification, is proposed. The adaptive algorithm used for updating parameters is kept the <WP=90>same as the one used by the standard MRAC system. Performance analysis of the proposed adaptive control scheme is verified based on the mean square tracking error criterion and thetracking error theory. Finally, the adaptive controllers with and without fixed compensator are compared and applied to an aircraft更多还原