双线性系统多目标控制问题研究

研究了双线性系统的多目标控制问题. 首先把多目标控制问题, 通过效用函数技术转化为一个单目标最优控制问题, 其中, 效用函数是多个二次型性能指标的非线性函数, 因此, 在动态规划的意义下是不可分的. 然后, 为了克服不可分对求解带来的困难, 提出了一种两级最优控制算法. 下级用动态规划求解一个参数化的具有双线性---二次型结构的辅助 Lagrangian 问题；上级迭代调整辅助 Lagrangian 问题中的参数向量. 不断重复这个过程, 直至最优性条件被满足.     

多种燃料锅炉效率曲线的辨识方法

多种燃料锅炉的运行优化是钢铁联合企业节约能源的一个主要方法,但是这种锅炉的效率曲线的辨识却是一个复杂而困难的非线性最小二乘问题。借鉴大系统递阶算法的思想,本文提出了一种新的两级递阶辨识的算法。这种算法通过预估关联量,将非线性最小二乘问题转化为两级线性最小二乘问题。实验的结果证明:这种算法是一种有效的算法。     

受扰非线性离散系统的前馈反馈最优控制

利用逐次逼近法研究含外部扰动的非线性离散系统的线性二次型前馈反馈最优控制问题.首先将系统的最优控制问题转化为非线性两点边值问题族.其次,构造了该问题族的由精确线性项和非线性补偿项组成的解序列,并证明了解序列一致收敛到系统的最优解.最后,通过截取最优控制序列解中非线性补偿项的有限项,得到系统的前馈反馈次优控制(FFSOC)律及设计算法.仿真算例表明,该算法容易实现,且对抑制外部扰动的鲁棒性优于经典的反馈次优控制(FSOC).     

基于神经网络旋转二级倒立摆系统的预测控制

针对倒立摆的研究,提出了一种应用神经网络和预测控制相结合的算法,用于控制旋转二级倒立摆系统.为了提高跟踪精度和快速性,可以把线性控制的条件作为非线性最优控制性能指标的约束条件,并将一种新型的非线性混沌映射引入到神经网络的参数学习算法中实现其权值调节,以逐步建立被控对象合理的多步预测模型.仿真结果表明,神经网络预测控制算法具有响应速度快、控制效果好和跟踪精度高等特点.可使非线性系统的预测问题得到较好的解决.     

非线性相似组合大系统最优控制的逐次逼近过程

研究一类仿射非线性相似组合大系统关于二次型性能指标的最优控制问题．首先通过模型简化,将非线性相似组合大系统化为若干个准解耦的子系统;然后利用非线性系统最优控制的逐次逼近设计方法,将求解高阶强耦合的非线性两点边值问题简化为求解一族解耦的线性两点边值问题序列．该线性两点边值问题序列的解一致收敛于非线性相似组合大系统的最优控制,得到的最优控制律由线性最优控制的解析项与非线性补偿序列的极限项组成．通过截取最优控制非线性补偿序列的有限次逼近值．得到了非线性组合大系统的次优控制律．     

一类控制受约束非线性系统的基于单网络贪婪迭代DHP算法的近似最优镇定

提出一种贪婪迭代DHP (Dual heuristic programming)算法, 解决了一类控制受约束非线性系统的近似最优镇定问题. 针对系统的控制约束, 首先引入一个非二次泛函把约束问题转换为无约束问题, 然后基于协状态函数提出一种贪婪迭代DHP算法以求解系统的HJB (Hamilton-Jacobi-Bellman)方程. 在算法的每个迭代步, 利用一个神经网络来近似系统的协状态函数, 而后根据协状态函数直接计算系统的最优控制策略, 从而消除了常规近似动态规划方法中的控制网络. 最后通过两个仿真例子证明了本文提出的最优控制方案的有效性和可行性.     

超混沌系统同步非线性反馈控制

在将超混沌系统的控制、同步及反同步问题统一处理的基础上,给出了一种可实现超混沌系统同步的非线性状态反馈方法。该方法以非线性系统的线性化方法和极点配置理论为基础,把控制律分解为非线性和线性两部分之和,使超混沌系统得以控制、同步和反同步。对超混沌Newton-Leipnik系统的仿真实验表明了控制策略的有效性。     

基于神经网络误差修正的混沌控制与同步

针对混沌系统非线性强、多变量耦合等特点,提出了一种基于神经网络误差修正的自适应多变量混沌系统的广义预测控制算法,用线性广义预测控制器控制混沌系统,用神经网络对模型预测误差进行修正。算法中辩识过程模型用递推最小二乘法(RLS)、神经网络权值用Davidon最小二乘法(DLS)训练。这种算法对被控混沌系统的先验知识要求较少,无需知道被控系统的精确模型,数值仿真显示可实现混沌系统的宽范围控制与同步。     

线性时滞系统前馈-反馈次优控制: Taylor 级数法

研究线性时滞系统最优控制的前馈反馈近似设计问题.基于Taylor级数法,将系统的二次型最优控制问题转化为线性代数方程组的求解问题,给出了系统前馈反馈次优控制律的存在唯一性条件和Taylor级数表示形式.仿真算例验证了方法的有效性.     

一种非线性离散动态投入产出系统的最优控制

阐述了基于动态投入产出模型的最优控制理论,并对当前国内外的研究成果进行了对比研究。在分析其优缺点的同时,针对非线性离散动态投入产出系统的特点,提出了一种动态投入产出系统最优控制的逐次逼近方法。此方法首先将系统的最优控制问题转化为非线性两点边值问题族,然后通过构造线性两点边值问题族,将非线性两点边值问题转化为非齐次线性两点边值问题族;得到的最优控制律由精确控制项和非线性补偿项两部分组成,精确控制项可以通过求解Riccati方程求出其精确解,非线性补偿项由逐次逼近法求解一族线性伴随向量方程的解序列求得;最优控制律的最终目标是在规划期内使实际产出尽可能地与理想产出接近。实验仿真测试表明,采用逐次逼近法获得了非线性离散动态投入产出的最优系统控制,从而为最优控制问题的有效解决提供了参考和借鉴。     

Functional link artificial neural network applied to active noise control of a mixture of tonal and chaotic noise

Many practical noises emanating from rotating machines with blades generate a mixture of tonal and the chaotic noise. The tonal component is related to the rotational speed of the machine and the chaotic component is related to the interaction of the blades with air. An active noise controller (ANC) with either linear algorithm like filtered-X least mean square (FXLMS) or nonlinear control algorithm like functional link artificial neural network (FLANN) or Volterra filtered-X LMS (VFXLMS) algorithm shows sub-optimal performance when the complete noise is used as reference signal to a single controller. However, if the tonal and the chaotic noise components are separated and separately sent to individual controller with tonal to a linear controller and chaotic to a nonlinear controller, the noise canceling performance is improved. This type of controller is termed as hybrid controller. In this paper, the separation of tonal and the chaotic signal is done by an adaptive waveform synthesis method and the antinoise of tonal component is produced by another waveform synthesizer. The adaptively separated chaotic signal is fed to a nonlinear controller using FLANN or Volterra filter to generate the antinoise of the chaotic part of the noise. Since chaotic noise is a nonlinear deterministic noise, the proposed hybrid algorithm with FLANN based controller shows better performance compared to the recently proposed linear hybrid controller. A number of computer simulation results with single and multitone frequencies and different types of chaotic noise such as logistic and Henon map are presented in the paper. The proposed FLANN based hybrid algorithm was shown to be performing the best among many previously proposed algorithms for all these noise cases including recorded noise signal.     

基于神经网络的非线性系统多步预测控制

针对离散非线性系统,利用非线性激励函数的局部线性表示,提出一种可用于非线性过程的神经网络多步预测控制方法,并给出了控制律的收敛性分析．该方法将非线性系统处理成简单的线性和非线性两部分,对复杂的非线性多步预测方程给出了直观而有效的线性形式,并用线性预测控制方法求得控制律,避免了复杂的非线性优化求解．仿真结果表明了该算法的有效性．     

Control of chaotic dynamical systems using radial basis function network approximators

This paper presents a general control method based on radial basis function networks (RBFNs) for chaotic dynamical systems. For many chaotic systems that can be decomposed into a sum of a linear and a nonlinear part, under some mild conditions the RBFN can be used to well approximate the nonlinear part of the system dynamics. The resulting system is then dominated by the linear part, with some small or weak residual nonlinearities due to the RBFN approximation errors. Thus, a simple linear state-feedback controller can be devised, to drive the system response to a desirable set-point. In addition to some theoretical analysis, computer simulations on two representative continuous-time chaotic systems (the Duffing and the Lorenz systems) are presented to demonstrate the effectiveness of the proposed method.     

混沌系统的RBF神经网络非线性补偿控制

设计RBF神经网络非线性补偿控制器,提出了混沌系统线性状态反馈的复合控制方法,将可调系统混沌行为镇定到期望目标位置或者变成周期运动.用Lorenz方程作仿真实验,结果证明了该方法的有效性.     

Control and simulation for a closed-chain dual redundant ' manipulator system

The two-level hierarchical control scheme is adopted to solve a dual-chain robotic system formed by two redundant manipulators grasping a common object with hard point contacts. The upper-level coordinator gathers all the necessary information to resolve the force distribution problem so as to generate all the desired contact forces With all the desired contact forces being specified, the dynamics of each chain are decoupled. Therefore, the lower-level local control problem can be treated as a general open-chain redundant manipulator problem. Furthermore, the Compact-Dual LP method is applied to resolve the upper-level coordination problem; while the Compact QP method associated with the computed torque control method is adopted to solve the redundant manipulator problem. To obtain proper simulation results, the simulated actual contact forces are formulated and the corresponding simulation problem of the closed-chain robotic system is also completely solved in the article. Simulation results show that the two-level hierarchical control scheme is an effective and efficient algorithm for resolving the large-scale control problem of multiple-chain robotic systems. © 1995 John Wiley & Sons, Inc.     

基于D-FNN的混沌铁磁谐振系统非线性补偿控制方法

针对电网混沌铁磁谐振系统产生的混沌现象,提出基于动态模糊神经网络的混沌铁磁谐振系统非线性补偿控制方法。该方法采用动态模糊神经网络来逼近系统的非线性部分,消除了过电压的混沌现象,将系统稳定到目标位置,实现了对系统的非线性补偿控制。Matlab仿真结果表明,基于动态模糊神经网络的非线性补偿控制方法控制结果正确,响应快速。     

基于自然梯度的后非线性盲解卷算法

给出了一个基于自然梯度的后非线性多信道盲解卷算法。混合系统由线性卷积混合后接一个可逆非线性失真组成。分离系统由多层感知器(非线性部分)后接一个线性盲解卷过程(线性部分)组成。分离系统的线性部分和非线性部分参数学习都采用自然梯度算法。仿真结果显示,自然梯度算法比传统梯度算法收敛速度更快,分离效果更好。     

基于双线性模型的动态系统优化和参数估计集成方法

针对双线性模型与实际系统之间的差异，提出一种基于双线性模型求解非线性动态系统最优控制的迭代算法。该算法通过重复求解修正的基于双线性模型的优化控制问题和参数估计问题，获得实际系统的最优解。同时提出求解修正的基于双线性模型的优化控制问题的一种新的分解方法，克服了非线性和双线性两点边值问题求解的困难。仿真例子表明该算法的有效性和实用性。     

A rapid learning and dynamic stepwise updating algorithm for flatneural networks and the application to time-series prediction

A fast learning algorithm is proposed to find an optimal weights of the flat neural networks (especially, the functional-link network). Although the flat networks are used for nonlinear function approximation, they can be formulated as linear systems. Thus, the weights of the networks can be solved easily using a linear least-square method. This formulation makes it easier to update the weights instantly for both a new added pattern and a new added enhancement node. A dynamic stepwise updating algorithm is proposed to update the weights of the system on-the-fly. The model is tested on several time-series data including an infrared laser data set, a chaotic time-series, a monthly flour price data set, and a nonlinear system identification problem. The simulation results are compared to existing models in which more complex architectures and more costly training are needed. The results indicate that the proposed model is very attractive to real-time processes.     

Application of stochastic control techniques to chaotic nonlinearsystems

A control algorithm based on stochastic control techniques is devised for chaotic nonlinear systems. The algorithm uses a state estimator based on the Kalman filter, and yields performance improvements in at least some regions of state space with respect to that obtainable by use of a controller utilizing only the conditional mean of the system state vector. The method is applied to two typical chaotic nonlinear systems (the Henon-Heiles system and the Lorenz system), and their behavior with control is explored numerically