ROBUST FAILURE DETECTION AND ISOLATION IN NON-LINEAR CONTROL SYSTEMS

Robust Failure Detection and Isolation in Non-Linear Control Systems

This paper is concerned with an adaptive Sliding Mode Control (SMC) for non-linear discrete-time systems with mismatched uncertainty. The proposed adaptive SMC is composed of the equivalent and the switching controls, and it is constructed on the basis of the information from the proposed estimator to take the estimates for the state and uncertainty. It is assumed that some of the state variables can be measured and the uncertainty in the discrete-time systems is parameterised linearly in the unknown constant terms. The estimates for the state and the uncertainty are taken by applying the proposed weighted least squares method. It is verified based on the Lyapunov scheme according to which the estimation errors converge to zero as time increases, and the trajectory of the system response is stable as the sliding surface is suitably selected. The effectiveness of the proposed adaptive SMC is illustrated by the simulation experiment in a simple numerical example.

The problem of dynamic output feedback equivalence of nonlinear observable discrete-time control systems with outputs has been studied by Pawluszewics-Bartosiewics. In this paper we use those results to achieve equivalence of a nonlinear system to a linear one. In this way we obtain an algebraic criterion of dynamic output linearization for discrete-time systems. We recall the concept of anticipating output equivalence, where transformations of outputs and controls depend on future values, and the concept of dynamic output equivalence, depending on past and present controls and outputs values. In order to give the criterion of linearisation of a discrete-time control system we have to use the difference output algebra and the difference output universe associated with the given system.

摘要: 确定学习运用自适应控制和动力学系统的概念与方法, 研究未知动态环境下的知识获取、表达、存储和利用等问题. 针对产生周期或回归轨迹的连续 非线性动态系统, 确定学习可以对其未知系统动态进行局部准确建模, 其基本要 素包括: 1)使用径向基函数(Radial basis function, RBF)神经网络; 2)对于周期(或回归)状态轨迹 满足部分持续激励条件; 3)在周期(或回归)轨迹的邻域内实现对非线性系统动态的局部准确神经网络逼近(局部准确建模); 4)所学的知识以时不变且空间分布的方式表达、以常值神经网络权值的方式存储, 并可在动态环境下用于动态模式的快速识别或者闭环神经网络控制. 本文针对离散动态系统, 扩展了确定学习理论, 提出一个根据时态数据序列对离散动态系统进行建模与控制的框架. 首先, 运用确定学习原理和离散系统的自适应辨识方法, 实现对产生时态数据的离散非线性系统的未知动态进行局部准确的神经网络建模, 并利用此建模结果对时态数据序列进行时不变表达. 其次, 提出时态数据序列的基于动力学的相似性定义, 以及对离散动态系统产生的时态数据序列(亦可称为动态模式)进行快速识别方法. 最后, 针对离散非线性控制系统, 实现了基于时态数据序列对控制系统动态的闭环辨识(局部准确建模). 所学关于闭环动态的知识可用于基于模式的智能控制. 本文表明确定学习可以为时态数据挖掘的研究提供新的途径, 并为基于数据的建模与控制等问题提供新的研究思路. 关键词: 确定学习 / 时态数据序列 / 离散动态系统 / 基于数据的建模 / 部分持续激励条件 / 时态数据挖掘 / 动态模式识别 / 基于模式的控制

Neural-network-based accelerated safe Q-learning for optimal control of discrete-time nonlinear systems with state constraints.

Discrete-time neural network output feedback control of nonlinear discrete-time systems in non-strict form

In this work we study the design and analysis for nonlinear discrete time active control system with boring charter. It is shown that most analysis result for continuous time nonlinear system can be extended to the discrete time case. In previous studies, a method of nonlinear Model Following Control System (MFCS) was proposed by Okubo (1985). In this study, the method of nonlinear MFCS will be extended to nonlinear discrete time system with boring charter. Nonlinear systems which are dealt in this study have the property of norm constraints ║ƒ (v (k))║≤α + β║v (k)║^γ , where α≥0, β≥0, 0≤γ≤1. When 0≤γ<1. It is easy to extend the method to discrete time systems. But in the case γ = 1 discrete time systems, the proof becomes difficult. In this case, a new criterion is proposed to ensure that internal states are stable. We expect that this method will provide a useful tool in areas related to stability analysis and design for nonlinear discrete time systems as well.

In this paper, a learning-based predictive control (LPC) scheme is proposed for adaptive optimal control of discrete-time nonlinear systems under stochastic disturbances. The proposed LPC scheme is different from conventional model predictive control (MPC), which uses open-loop optimization or simplified closed-loop optimal control techniques in each horizon. In LPC, the control task in each horizon is formulated as a closed-loop nonlinear optimal control problem and a finite-horizon iterative reinforcement learning (RL) algorithm is developed to obtain the closed-loop optimal/suboptimal solutions. Therefore, in LPC, RL and adaptive dynamic programming (ADP) are used as a new class of closed-loop learning-based optimization techniques for nonlinear predictive control with stochastic disturbances. Moreover, LPC also decomposes the infinite-horizon optimal control problem in previous RL and ADP methods into a series of finite horizon problems, so that the computational costs are reduced and the learning efficiency can be improved. Convergence of the finite-horizon iterative RL algorithm in each prediction horizon and the Lyapunov stability of the closed-loop control system are proved. Moreover, by using successive policy updates between adjoint time horizons, LPC also has lower computational costs than conventional MPC which has independent optimization procedures between two different prediction horizons. Simulation results illustrate that compared with conventional nonlinear MPC as well as ADP, the proposed LPC scheme can obtain a better performance both in terms of policy optimality and computational efficiency.

This article addresses the robust stability for a class of nonlinear uncertain discrete-time systems with convex polytopic of uncertainties. The system to be considered is subject to both interval time-varying delays and convex polytopic-type uncertainties. Based on the augmented parameter-dependent Lyapunov-Krasovskii functional, new delay-dependent conditions for the robust stability are established in terms of linear matrix inequalities. An application to robust stabilization of nonlinear uncertain discrete-time control systems is given. Numerical examples are included to illustrate the effectiveness of our results. MSC:15A09, 52A10, 74M05, 93D05.

In this paper, the output feedback based finite-horizon near optimal regulation of nonlinear affine discrete-time systems with unknown system dynamics is considered by using neural networks (NNs) to approximate Hamilton-Jacobi-Bellman (HJB) equation solution. First, a NN-based Luenberger observer is proposed to reconstruct both the system states and the control coefficient matrix. Next, reinforcement learning methodology with actor-critic structure is utilized to approximate the time-varying solution, referred to as the value function, of the HJB equation by using a NN. To properly satisfy the terminal constraint, a new error term is defined and incorporated in the NN update law so that the terminal constraint error is also minimized over time. The NN with constant weights and time-dependent activation function is employed to approximate the time-varying value function which is subsequently utilized to generate the finite-horizon near optimal control policy due to NN reconstruction errors. The proposed scheme functions in a forward-in-time manner without offline training phase. Lyapunov analysis is used to investigate the stability of the overall closed-loop system. Simulation results are given to show the effectiveness and feasibility of the proposed method.

Event-triggered controller design of nonlinear discrete-time networked control systems in T-S fuzzy model

A novel tracking control algorithm based on sliding mode prediction is proposed for a class of discrete-time nonlinear uncertain systems in this paper. By constructing a sliding mode prediction model at first, the future, present and history information of the system can be used to improve the performance of sliding mode control. Because of sliding mode reference trajectory, the reachability of sliding mode is achieved, and the reaching mode can be determined by designers in advance. Due to feedback correction and receding horizon optimization, the influence of uncertainty can be compensated in time and control law is obtained subsequently. Theoretical analysis proves the closed-loop system possesses robustness to matched or unmatched uncertainty, without requiring the known bound of uncertainty. Simulation results illustrate the validity of the presented algorithm.

The problem of finite-time reliable guaranteed cost fuzzy control for discrete-time nonlinear systems with actuator faults is investigated in this article. We first provide a novel and explicit interpretation for finite-time reliable guaranteed cost fuzzy control. Second, a suitable reliable controller is designed such that the provided performance criterion is satisfied for the normal and possible actuator fault cases. Moreover, the switched Lyapunov function approach is utilised for relaxing the conservatism of these obtained conditions and the related optimisation problem is also offered to minimise the guaranteed cost performance bound. Finally, an illustrative example is given to show the validity of the proposed scheme.

An algorithm for control of nonlinear discrete-time systems is presented in the paper. Controlled system is linearized by dynamic output feedback so that the linearized closed loop system is equivalent to a predefined discrete-time transfer function representing reference model of the control system. Choice of the reference model provides placement of zeroes and poles of the closed loop system. In the proposed approach at least one of the poles is not fixed and moves during the time of transient process. Evolution of the pole occurs according to certain rules formalized and implemented in the algorithm by means of fuzzy logic. Therefore, the parameters of the transfer function of the linearized closed loop system may be understood as nonlinear functions which depend on the current control error and its derivative. Thus the poles of the closed loop system are placed dynamically according to the predefined rules providing necessary behavior of the control system. Controlled system has to be represented by a nonlinear model with no couplings between different time instances what can be performed by training an Artificial Neural Network of the specific structure. The developed theory and control algorithm are illustrated by means of numerical example.

The paper focuses on unification of the study of continuous- and discrete-time nonlinear control systems using the mathematical tools of pseudo-linear algebra. The realization problem is studied through the language of differential forms. Necessary and sufficient conditions are given for the existence of the state space realization of the nonlinear input-output equation defined in terms of the pseudo-linear operator. The sufficiency part of the proof gives a constructive procedure (up to integration of the one-forms) for obtaining the observable state equations. The special cases for continuous- and discrete-time systems, described either in terms of the shift or difference operator, follow from the general result as the special cases. Moreover, the result accommodates also the systems described via q-shift or q-difference operators.