基于神经网络Smith预估器的预测控制

针对非线性系统时滞问题,给出了一种新型的单神经元Smith预测控制算法.神经网络的预测控制器由不完全微分的单神经元自适应PID控制器和神经网络的Smith预估器组成.预估器对输出进行多步预测,控制器超前动作以消除时滞对系统的影响.不完全微分的单神经元自适应PID控制器通过改进的Hebb学习规则实现其权值调节,通过权系数的在线调整实现自适应控制.仿真实验证明了该方法具有较快的响应速度和较好的响应性能.     

多时滞多变量系统的自校正控制及其应用

本文提出一种具有k步增量预估器的多变量自校正控制算法,使其对负荷扰动有较强的 克服能力;并引入辅助控制作用,使算法适用于多时滞系统.该算法用于造纸机的纸张定量和 水份控制,结果令人满意.     

三角域上双变量Chebyshev 多项式及其与Bernstein 基的转换

为了更好的解决三角域上的Bézier 曲面在CAGD 中的最佳一致逼近问题, 构造出了三角域上的双变量Chebyshev 正交多项式,研究了与单变量Chebyshev 多项式相类 似的性质,并且给出了三角域上双变量Chebyshev 基和Bernstein 基的相互转换矩阵。通过 实例比较双变量Chebyshev 多项式与双变量Bernstein 多项式以及双变量Jacobi 多项式的最 小零偏差的大小,阐述了双变量Chebyshev 多项式的最小零偏差性。     

比例积分型广义预测控制系统的稳定性分析

广义预测控制(GPC)系统的闭环稳定性一直是控制理论分析的难点．本文通过对预测控制系统闭环特征多项式的研究,证明了一步预测比例积分型GPC系统的闭环稳定性;同时,利用根轨迹法分析了控制参数与闭环极点的关系,明确了比例积分型GPC中参数的物理意义．通过一个数值仿真例子,从频域分析的角度说明了比例积分型GPC较普通GPC的优越性．     

基于状态空间的有约束稳定广义预测控制

针对一类有约束的稳定广义预测控制问题,提出一种基于状态空间的稳定广义预测控制算法。首先通过传递函数的状态空间实现,得到被控对象的离散状态空间形式;然后引入Deadbeat状态反馈矩阵并给出约束条件的等价性定理,实现了约束条件的等价转化;最后通过等价约束条件优化性能指标函数求解控制律。仿真实例表明该方法具有良好的稳定性。     

多变量解耦预估自校正控制器及其应用

本文在 Chien 的工作基础上提出了一种多变量解耦预估广义最小方差自校正控制器.该算法可克服变量间的干扰,实现静、动态解耦并提高系统的稳定性.由于同时引入了对角矩阵解耦法及 Smith 预估器,自校正控制器的设计过程可简化为单变量无时延系统.仿真研究与实际应用表明:本算法响应速度快、超调小,结果是令人满意的.     

基于抽象解释的服务间消息的数据约减

面向服务软件中服务间消息的变量值可能存在无穷域的情况,从而导致模型检测时产生状态空间爆炸问题。为了使终止性验证在实践上可行,需要约减模型状态空间的大小,使得计算时间和空间需求合理。为此,基于抽象解释的区间抽象理论扩展了经典区间抽象域方法,并在统一的区间抽象域方法上借助异常控制流图对变量进行区间分析,在此基础上逆向分析得到服务间消息的变量区间集。变量区间上任意值相对于终止性验证是等价性,因此从每一个变量区间集中选取一个代表值,可组成服务间消息变量的约减值,从而为异常处理的终止性验证提供了约减的初始配置,有效避免了状态空间爆炸。     

多变量加权多步预报控制*

对于线性多变量系统,本文给出了一种完全不同于以往的全状态反馈或观测器-控制器型的算法——多变量加权多步预报控制(MWLPC)算法。这种算法除引进了预测控制中的多步输出预报、滚动优化等机制外,最重要的是在二次型性能指标中引入了可调的多项式或有理分式矩阵权因子;适当选取这些权因子,便可按设计要求、仅用系统的输出信息反馈便能任意配置闭环系统的特征矩阵,从而保证闭环稳定性和其他优良性质。此外,该算法不改变原系统的零点,因而适用于非最小相位系统。     

一种多变量连续时间预测控制方法

将积分作用自然引入单变量连续时间域预测控制规律,并通过结合多变量频域设计方 法--特征轨迹法,将单变量预测控制推广到多变量情况.仿真结果表明,该多变量算法是有 效的.     

非线性广义最小方差控制律综述

如何设计简单的控制策略对复杂非线性系统进行控制是控制界还未解决的难题．非线性广义最小方差控制律的提出使得非线性控制器的设计可以基于更为一般的非线性模型，并且控制器易于实现．整个系统包含时滞环节，稳定的非线性输入子系统和一个可以用多项式或者状态空间描述的子系统．通过最小化由误差加权项、状态加权项和输入加权项组成的信号的方差得到优化控制器．在系统为开环稳定的情况下，可用史密斯预估器进行控制．本文首先介绍了非线性广义最小方差控制的发展进程，然后综述了基于状态空间和多项式描述的系统的非线性广义最小方差控制器的设计以及其现状和今后的发展方向．     

A state-space description for GPC controllers

Generalized predictive control (GPC)-type control algorithms traditionally derived in the polynomial domain are derived in this paper in the state-space domain, but following the polynomial approach due to Clarke et al. (1987). Relations between the polynomial and state-space parameters are presented. Some possible state-space representations which were used earlier in different publications are discussed. The problem of deriving the GPC algorithm in the state-space domain is solved for the unrestricted case as well as for the case of restricted control and output horizons. Some properties of the state estimate for this problem are presented; in particular, two methods of Kalman filtering—optimal and asymptotic—are proposed. The solution is valid for any possible (minimal or non-minimal) state-space representation. Another approach to this problem is by the ‘dynamic programming method’ and solving the Riccati equation (Bitmead et al. 1990). This approach is also presented in this paper but the method differs from this earlier work and does not require extending the state dimension. Ultimately, certain features of the state-space approach are discussed, such as (a) the opportunity for straightforward analysis of the transient states produced by switching on the regulator, by changing the set-point or by changing the regulator parameters; (b) easy extension to the multidimensional case; and (c) the possibility of introducing nonlinearities into the model     

A combination of fuzzy logic and neural network controller for multiple-input multiple-output systems

Generally, the difficulty of multiple-input multiple-output (MIMO) systems control is how to overcome the coupling effects between the degrees of freedom. Owing to the computational burden and dynamic uncertainty of MIMO systems, the model-based decoupling approach is not practical for real-time control. A hybrid fuzzy logic and neural network controller (HFNC) is proposed here to overcome this problem and to improve the control performance. Firstly, a traditional fuzzy controller (TFC) is designed from a single-input single-output (SISO) systems viewpoint for controlling the degrees of freedom of a MIMO system. Secondly, an appropriate coupling neural network controller is introduced into the TFC for compensating the system coupling effects. This control strategy not only can simplify the implementation problem of fuzzy control but also can improve the control performance. The state-space approach for fuzzy control systems stability analysis is employed to evaluate the stability and robustness of this intelligent hybrid controller. In addition, a dynamic absorber with a twolevel mass-spring-damper structure was designed and constructed to verify the stability and robustness of a HFNC by numerical simulation and to investigate the control performance by comparing the experimental results of the HFNC with that of a TFC for this MIMO system.     

A model predictive control approach with relevant identification in dynamic PLS framework

In this paper, a generalized predictive control (GPC) scheme under a dynamic partial least squares (PLS) framework is proposed. At the modeling stage, a model predictive control relevant identification (MRI) approach is used to improve the identification of the model. Within PLS framework, the MIMO system can be automatically decomposed into several SISO subsystems in the latent space. For each subsystem, MRI is implemented and GPC is designed independently. With the advantage of MRI and PLS framework, fewer parameters are needed to be estimated in the identification stage, nonsquare and ill-conditioned system can be handled naturally, control parameters tuning is easier and better control performance can be obtained. Furthermore, the computing time of control action which is very crucial for GPC on-line application decreases since each GPC is running in SISO subsystem in parallel. The results of two simulation examples and a laboratory experiment demonstrate the merit of the proposed method.     

Block transformations of a class of multivariable control systems to four basic block companion forms

Various block transformations are presented for transforming a class of MIMO state equations in general coordinates to four basic block companion forms so that the analysis and control system design of a MIMO system in the time and frequency domains can easily be performed, and the classical lines of thought for SISO systems can be extended to MIMO systems. The invariant structure and the invariant characteristic matrix polynomial of four basic block companion forms are investigated.     

State-space realizations of MIMO 2D discrete linear systems Elementary operation and variable inversion approach

Elementary operations are an alternative to the polynomial system matrix for studying basic systems theoretic properties of standard (1D) linear systems, such as state-space realizations. In the general area of nD systems, there is a basic need to analyse both rational and polynomial matrices in n indeterminates. This paper develops key results on using elementary operations and variable inversions to produce equivalent state-space realizations of MIMO 2D systems.     

A novel neural internal model control for multi-input multi-output nonlinear discrete-time processes

An internal model-based neural network control is proposed for unknown non-affine discrete-time multi-input multi-output (MIMO) processes in nonlinear state space form under model mismatch and disturbances. Based on the neural state-space model built for an unknown nonlinear MIMO state space process, an approximate internal model and approximate decoupling controllers are derived simultaneously. Thus, the learning of the inverse process dynamics is not required. A neural network model-based extended Kalman observer is used to estimate the states of a nonlinear process as not all states are accessible. The proposed neural internal model control can work for open-loop unstable processes with its closed-loop stability derived analytically. The application to a distributed thermal process shows the effectiveness of the proposed approach for suppressing nonlinear coupling and external disturbances and its feasibility for the control of unknown non-affine nonlinear discrete-time MIMO state space processes.     

An approximate method for numerically solving fractional order optimal control problems of general form

In this article, we discuss fractional order optimal control problems (FOCPs) and their solutions by means of rational approximation. The methodology developed here allows us to solve a very large class of FOCPs (linear/nonlinear, time-invariant/time-variant, SISO/MIMO, state/input constrained, free terminal conditions etc.) by converting them into a general, rational form of optimal control problem (OCP). The fractional differentiation operator used in the FOCP is approximated using Oustaloup’s approximation into a state-space realization form. The original problem is then reformulated to fit the definition used in general-purpose optimal control problem (OCP) solvers such as RIOTS_95, a solver created as a Matlab toolbox. Illustrative examples from the literature are reproduced to demonstrate the effectiveness of the proposed methodology and a free final time OCP is also solved.     

Frequency-domain stability criteria for SISO and MIMO nonlinear feedback systems with constant and variable time-delays

New frequency-domain criteria are proposed for the $L_2$-stability of both nonlinear single-input-single-output (SISO) and nonlinear multiple-input-multiple-output (MIMO) feedback systems, described by nonlinear integral equations. For SISO systems, the feedback block is a constant scalar gain in product with a linear combination of first-and-third-quadrant scalar nonlinearities (FATQNs) with time-delay argument functions; and, for MIMO systems, it is a constant matrix gain in product with a linear combination of vector FATQNs also with time-delay argument functions. In both the cases, the delay function in the arguments of the nonlinearities may be, in general, i) zero, ii) a constant, iii) variable-time and iv) fixed-history (only for SISO systems). The stability criteria are derived from certain recently introduced algebraic inequalities concerning the scalar and vector nonlinearities, and involve the causal+anticausal O''Shea-Zames-Falb multiplier function (scalar for SISO systems and matrix for MIMO systems). Its time-domain $L_1$-norm is constrained by the coefficients and characteristic parameters (CPs) of the nonlinearities and, in the case of the time-varying delay, by its rate of variation also. The stability criteria, which are independent of Lyapunov-Krasovskii or Lyapunov-Razumikhin functions and do not seem to be derivable by invoking linear matrix inequalities, seem to be the first of their kind. Two numerical examples for each of SISO and MIMO systems illustrate the criteria.     

Operator-Based Robust Nonlinear Control for SISO and MIMO Nonlinear Systems With PI Hysteresis

In this paper, operator based robust nonlinear control for single-input single-output (SISO) and multi-input multi-output (MIMO) nonlinear uncertain systems preceded by generalized Prandtl-Ishlinskii (PI) hysteresis is considered respectively. In detail, by using operator based robust right coprime factorization approach, the control system design structures including feedforward and feedback controllers for both SISO and MIMO nonlinear uncertain systems are given, respectively. In which, the controller design includes the information of PI hysteresis and its inverse, and some sufficient conditions for the controllers in both SISO and MIMO systems should be satisfied are also derived respectively. Based on the proposed conditions, influence from hysteresis is rejected, the systems are robustly stable and output tracking performance can be realized. Finally, the effectiveness of the proposed method is confirmed by numerical simulations.      

Generalized predictive control based on neural networks

This paper presents the Generalized Predictive Control (GPC) strategy based on Artificial Neural Network (ANN) plant model. To obtain the step and the free process responses which are needed in the generalized predictive control strategy we iteratively use a multilayer feedforward ANN as a one-step-ahead predictor. A bioprocess was chosen as a realistic nonlinear SISO system to demonstrate the feasibility and the performance of this control scheme. A comparison was made between our approach and the adaptive GPC (AGPC).