【作者】 陈雪芹；

【导师】 张迎春； 耿云海；

【作者基本信息】 哈尔滨工业大学 ， 控制理论与控制工程， 2008， 博士

【摘要】 控制系统的集成故障诊断与容错控制作为故障诊断与容错控制的交叉领域,可以避免单独设计故障诊断或容错控制引起的时间延迟以及对故障诊断结果的完全依赖,能够为现代控制系统长寿命高可靠性运行提供有利保障。本文以卫星姿态控制系统为应用背景,设计三类集成故障诊断与容错控制算法,主要内容如下:对控制系统中的各种故障进行分类与建模,并根据内部资料中对一些卫星姿态控制系统测试与在轨运行中的故障的记载,对故障进行详细分析,设置本文研究的飞轮与陀螺故障类型与大小,构建全文统一的卫星姿态闭环系统仿真环境。闭环控制仿真结果也为验证后续进行的其它容错控制结果的有效性提供基准。研究基于有效性因子/LQR的集成故障诊断与容错控制方法。分别针对敏感器故障、执行机构故障、敏感器/执行机构故障三种情况,利用偏差分离原理,相应地引入测量有效性因子、控制有效性因子、测量/控制有效性因子表示系统的故障程度。通过二阶Kalman滤波估计对有效性因子值以及状态变量进行估计,同时设计线性二次调节器（LQR）作为重构控制器。将状态估计结果作为原控制器和重构控制器的输入,通过统计假设检验进行判断,当确认故障发生时,引入重构容错控制器,与原控制器联合对故障状况下的控制系统进行补偿控制。当统计假设检验的结果说明系统无故障时,则不引入重构控制器。基于有效性因子的故障诊断与基于LQR的重构控制器联合设计实现了集成故障诊断与容错控制。研究基于IMM/EA的集成故障诊断与容错控制方法。利用随机混合系统描述故障系统,考虑执行机构和敏感器故障,根据系统所有可能会出现的故障的先验知识,确定N个系统模型逼近混合系统。当系统中部分或者全部状态发生突变时,采用交互式多模型（IMM）估计器得到故障发生的位置以及大小。通过重构机制判断,当确认故障发生时,进行容错控制器重构设计,与原控制器联合对故障状况下的控制系统进行补偿控制。利用故障模型中的动力学部分,采用输出反馈或（以及）状态反馈进行特征结构配置（EA）设计重构控制器。基于IMM的故障诊断与基于EA的重构控制器联合设计实现集成故障诊断与容错控制。针对集成故障诊断与容错控制中的鲁棒容错控制问题,设计基于混合H₂/H∞控制的鲁棒容错控制方法。考虑执行机构和敏感器故障,利用故障诊断的结果,并考虑故障诊断结果和系统参数的不确定性,分别采用H₂和H∞范数作为系统故障诊断和容错控制的性能指标,设计输出反馈混合H₂/H∞控制器,运用线性矩阵不等式（LMI）方法,通过给定的性能指标γ∞,调节故障诊断性能指标γ2,求解集成故障诊断与容错控制问题,并使得包含该集成故障诊断与容错控制器的闭环系统鲁棒性能最优。针对线性系统和非线性系统分别设计基于鲁棒模型预测控制（RMPC）和基于鲁棒自适应的鲁棒容错控制方法。RMPC控制将鲁棒概念和预测控制的滚动优化原理进行融合,通过在线更新系统模型和滚动优化,对具有输入和输出约束的线性不确定系统,同时考虑执行机构和传感器的故障以及FDD的不确定性,将不确定控制系统完整性设计转化为对新系统模型的鲁棒控制器设计问题。基于鲁棒自适应的容错控制引入在线估计,可以解决参数扰动引起的系统故障问题,利用自适应控制Lyapunov函数（ACLF）设计反馈,使系统实现自适应稳定,并采用逆最优控制方法求解。为降低鲁棒容错控制器的计算负担,首先通过姿态确定子系统对测量敏感器中的各种误差和故障,包括测量误差、噪声、漂移等,进行修正,得到修正后的姿态测量信息。然后,再由FDD子系统进行故障诊断,此时的FDD子系统主要是针对执行机构进行的故障诊断。最后,自适应逆最优控制器根据FDD得到的执行机构故障模型设计容错控制器,以满足故障状况下的系统性能要求。更多还原

【Abstract】 As an interdisciplinary domain between Fault Detection and Diagnosis （FDD） and Fault-Tolerant Control （FTC）, Integrated Fault Detection and Diagnosis and Fault-Tolerant Control of control system enable to avoid the time-delay induced by single FDD or FTC and the absolute dependence by the results of FDD to present a benefited insurance for the long-life and reliable circulation of modern control system. Regarding satellite attitude control system as the application background, three kinds of integrated FDD and FTC algorithm are developed in this dissertation. Here is the main content in detail:First, classifying and modeling the faults for control system. Combining several actual faults of satellite attitude control system recorded in internal materials, the thesis setups the category and range of flywheel and gyro faults and builds a uniform satellite attitude control system simulation environment which is utilized in the later chapters. What is more, two simple controllers are simulated in the satellite attitude closed-loop control system provided the benchmark to validate the efficiencies of other fault-tolerant controllers that are presented in later.Second, designing the Integrated Fault Detection and Diagnosis and Fault-Tolerant Control algorithm based on the effectiveness factor and Linear Quadric Regulator （LQR）. Focusing on the sensor faults, actuator faults and sensor/actuator faults, the dissertation applies measurement effectiveness factor, control effectiveness factor and measurement/control effectiveness factor respectively to express the range of system faults by using the bias-separated principle. Further, the two-stage kalman filter is used to obtain the effectiveness factor value and system estimated states are obtained. Meanwhile, a LQR controller is designed with the estimated results to accomplish the joint design of Fault Detection and Diagnosis and Fault-Tolerant Control.Third, designing the Integrated Fault Detection and Diagnosis and Fault-Tolerant Control algorithm based on IMM/EA. The faulty system is expressed by stochastic hybrid system and the location and range of faults can be determined by the IMM when the saltation occurs among the partial or whole states of the attitude estimation subsystem. Furthermore, using the dynamic part in the fault model and the output feedback or/and state feedback to assign the eigenstructure, a reconstructure controller is developed to compensate the original control system to complete the joint design of Fault Detection and Diagnosis and Fault-Tolerant Control.Finally, designing the Integrated Fault Detection and Diagnosis and Fault-Tolerant Control algorithm based on robust fault-tolerant control. The robust fault-tolerant control algorithms base on the robust model predictive control （RMPC） and H₂/H_∞are present respectively for the linear system and also a robust adaptive robust fault-tolerant control algorithm is also designed for the nonlinear system. Syncretizing the robust concept of H_∞control and the rolling optimal principle of predictive control, RMPC transforms the integrity of uncertainty control system to the problem of robust controller design for the new system model through the online updating of system model and rolling optimal. The robust adaptive fault-tolerant control enables to solve the system faults resulted by the parameter disturbances, where the system is adaptive stable by using feedback of the adaptive control Lyapunov function。更多还原