【作者】 许德智；

【导师】 李炜；

【作者基本信息】 兰州理工大学 ， 控制理论与控制工程， 2010， 硕士

【摘要】 本文以非线性系统为研究对象,基于逆系统方法研究了SISO系统、MIMO系统的主动容错控制问题,主要进行了以下几方面的研究工作：1)针对具有一定故障先验知识的非线性系统,为解决多模型主动容错控制在应对系统发生未知故障时控制律计算的快速性问题,提出了一种基于RBF逆系统方法的多模型内模主动容错控制方案。采用RBF对系统正常及各种先验故障情形的逆系统建模,并由此建立动态系统逆模型库,基于逆系统方法将逆模型与系统串联形成一伪线性系统,并对其设计了具有良好鲁棒性能的内模控制器。系统实际运行时,监控决策机制依据系统性能容忍度指标和模型失配度指标的时实计算分析,诊断系统所处运行模式,调用与之匹配的RBF逆模型,使系统始终通过模型切换与其逆模型的串联保持为不变的伪线性系统,从而在无需改变内模控制器的情况下,实现了对系统故障的主动容错。2)针对实际系统故障先验知识不易获取等问题,提出了一种基于RBF逆系统方法的故障诊断和调节的设计方法。通过设计基于非线性被控对象RBF逆模型的故障估计器来估计执行器故障,同时将RBF逆模型与被控对象串联成伪线性的复合系统,并引入内模控制。采用故障估计值作用于被控对象的逆系统,以产生相应的补偿,最终使被控对象和补偿后的逆系统串联仍能保持为不变的伪线性系统,从而在无需改变系统正常控制器参数的情形下,以逆系统补偿的方法达到容错控制的目的。3)针对实际系统存在不确定性及各种时变故障,将小样本建模方法LS-SVM引入用于系统的逆建模,借助数据驱动控制思想和无模型自适应控制方法,设计一个补偿控制器,用于系统参数摄动或发生故障时的逆模型的实时补偿,使系统在无需改变主控制参数情形下,以自适应逆补偿的方法,使得系统输出可以准确跟踪参考输出。4)针对状态空间形式的多变量非线性可逆系统,首先将逆系统方法引入,提出一种基于逆系统方法的多变量非线性系统故障诊断方法,通过设计非线性鲁棒滑模观测器实现状态估计,并结合逆系统方法,实现多变量非线性系统的执行器故障诊断。在此基础上又设计基于逆系统方法设计控制器,最终通过故障估计值对逆系统调节,实现了MIMO系统的主动容错控制。更多还原

【Abstract】 In this paper, as the study of nonlinear systems, active fault-tolerant control problems are studied based on inverse system method of the SISO system and MIMO system, mainly carried out research in the following areas:1) For certain priori knowledge fault of nonlinear systems, in order to solve the multi-model active fault-tolerant control system in response to an unknown fault of the fast control law calculation problem, proposes an active fault-tolerant control using multi-models based on Radial Basis Function (RBF) inverse system internal model methods. The method first uses RBF set up inverse system modeling for normal and faults with each kind of prior breakdown situation, then buildes inverse model bank for dynamic systems, based on inverse model of inverse system method connect the system become one pseudo-linear system, and designes with a good robust performance of the internal model controller. Actual run-time system, the decision-making mechanism of monitor is being calculated and analysis based on system performance indicators and model of tolerance mismatch index, The diagnosis system locates the movement pattern, call switching RBF inverse model, the system has always been seried with inverse model and remain unchanged pseudo-linear system, thus no need to change the situation of internal model controller for nonlinear systems and to achieve active fault-tolerant control purpose.2) For priori knowledge of the actual system faults and other problems is not easy to obtain, an approach of fault diagnosis and accommodation based on RBF inverse system method is designed. Through the design of RBF inverse model of nonlinear plant estimates actuator fault, while RBF inverse model connect the nonlinear plant become pseudo-linear system, and the introduction of internal model control. Estimates using fault acting on the inverse of the paint and to generate the corresponding compensation, and finally to the plant and the compensated inverse series can maintain the same pseudo-linear system, so no need to change the system in the normal control parameters of the case, in order to achieve the purpose of fault tolerant control through the inverse system method of compensation.3) For the real system with uncertainty and a variety of time-varying faults, the small sample of LS-SVM modeling method for systems is introducted using inverse modeling, using data-driven control idea and model free adaptive control method, design a compensation controller for real-time compensation when system parameter perturbation or fault occur, the system without changing the main controller parameters in the cases of adaptive inverse compensation method, allows the system output can accurately track the reference output.4) For the state space form of multi-variable nonlinear reversible systems, inverse system method is introducted, and a method based on inverse system of multi-variable nonlinear system fault diagnosis is proposed, through the design of nonlinear robust sliding mode observer device to achieve state estimation, and combined with inverse system method, to achieve the implementation of multi-variable nonlinear system fault diagnosis. On the basis of further design of the controller design based on inverse system method, the ultimate the inverse system is accommodated by the fault estimates, to achive MIMO system fault-tolerant control purpose.更多还原