Adaptive NN stabilization for stochastic systems with discrete and distributed time-varying delays

A new adaptive neural network(NN) output-feedback stabilization controller is investigated for a class of uncertain stochastic nonlinear strict-feedback systems with discrete and distributed time-varying delays and unknown nonlinear functions in both drift and diffusion terms.First,an extensional stability notion and the related criterion are introduced.Then,a nonlinear observer to estimate the unmeasurable states is designed,and a systematic backstepping procedure to design an adaptive NN output-feedback controller is proposed such that the closed-loop system is stable in probability.The effectiveness of the proposed control scheme is demonstrated via a numerical example.     

基于Kalman／UKF组合训练神经网络的初始对准方法

针对基于Unscented卡尔曼滤波(UKF)的神经网络训练学习方法存在的计算量大,实时性差的问题,提出了一种基于Kalman/UKF组合滤波原理的神经网络学习方法,该方法综合了Kalman滤波对线性系统和UKF对非线性系统的最优估计的优势,在保证神经网络权值估计精度的同时,有效降低了神经网络权值学习的计算量,提高了神经网络训练的实时性。最后将该利用方法训练的神经网络应用于惯性导航系统的非线性初始对准过程中,并进行了仿真研究。仿真结果表明利用提出的算法训练的神经网络与基于UKF训练的神经网络具有相同的对准精度和实时性,而提出的算法的有效降低了神经网络训练的计算量,提高了训练的运行效率,是解决惯性导航系统初始对准的一种有效和实用的方法。     

一种有约束的神经网络预测控制方法

针对非线性系统的控制问题，提出了一种采用黄金分割法的神经网络预测控制算法。该算法以神经网络作为预测模型，以黄金分割法用于优化控制器，其中以系统输入的约束条件作为黄金分割法的动态搜索区间。该算法解决了控制量的范围和变化速度受约束的情况下，未知非线性系统的预测控制问题，通过仿真研究证明了该算法计算速度、稳定性和杭扰动能力。     

Cooperative Neuro Adaptive Control of Leader Following Uncertain Multi-Agent Systems with Unknown Hysteresis and Dead-Zone

In this paper, a cooperative adaptive control of leader-following uncertain nonlinear multiagent systems is proposed. The communication network is weighted undirected graph with fixed topology. The uncertain nonlinear model for each agent is a higher-order integrator with unknown nonlinear functions, unknown disturbances and unknown input actuators. Meanwhile, the gains of input actuators are unknown nonlinear functions with unknown sign. Two most common behaviors of input actuators in practical applications are hysteresis and dead-zone. In this paper, backlash-like hysteresis and dead-zone are used to model the input actuators. Using universal approximation theorem proved for neural networks, the unknown nonlinear functions are tackled. The unknown weights of neural networks are derived by proposing appropriate adaptive laws. To cope with modeling errors and disturbances an adaptive robust structure is proposed. Considering Lyapunov synthesis approach not only all the adaptive laws are derived but also it is proved that the closed-loop network is cooperatively semi-globally uniformly ultimately bounded(CSUUB). In order to investigate the effectiveness of the proposed method, it is applied to agents modeled with highly nonlinear mathematical equations and inverted pendulums. Simulation results demonstrate the effectiveness and applicability of the proposed method in dealing with both numerical and practical multi-agent systems.     

一种新的小波网络快速学习混合策略

针对传统的小波网络梯度学习算法易于陷入局部极值、收敛速度慢且对初始参数很敏感的缺点,将全局性能优越的差异进化(DE)算法和最小二乘算法(LS)有机的结合起来,提出了一种新的快速学习混合策略。该混合学习算法思想是将待训练参数分为非线性和线性两类,利用差异进化算法对小波网络非参数进行全局优化训练,而最小二乘法用于快速训练网络连接权值。非线性函数逼近实验表明,小波网络逼近性能要远优于传统的BP神经网络,相对于使用随机梯度学习算法的小波网络,提出的混合学习算法收敛速度更快,且具有更小的均方差。     

基于HS-BP算法的尾矿库安全评价

为有效预防尾矿库事故的发生, 针对尾矿库事故率具有随机波动性和非线性的特点, 采用和声搜索算法(HSA)和BP神经网络建立尾矿库安全评价模型. 该方法利用HS算法对BP神经网络权值进行优化, 进而对尾矿库进行安全评价. 通过对辽宁本溪南芬尾矿库安全现状进行拟合预测, 结果表明:将HS算法和BP神经网络有机结合, 能够克服传统BP网络易陷入极小值、收敛速度慢得缺陷, 有效的刻画了尾矿库事故的随机波动特性, 并且预测能力均优于其他评价算法, 具有重要意义.     

Fault tolerant control based on stochastic distribution via RBF neural networks

A new fault tolerant control (FTC) via a controller reconfiguration approach for general stochastic nonlinear systems is studied. Different from the formulation of classical FTC methods, it is supposed that the measured information for the FTC is the probability density functions (PDFs) of the system output rather than its measured value. A radial basis functions (RBFs) neural network technique is proposed so that the output PDFs can be formulated in terms of the dynamic weighings of the RBFs neural network. As a result, the nonlinear FTC problem subject to dynamic relation between the input and the output PDFs can be transformed into a nonlinear FTC problem subject to dynamic relation between the control input and the weights of the RBFs neural network approximation to the output PDFs. The FTC design consists of two steps. The first step is fault detection and diagnosis (FDD), which can produce an alarm when there is a fault in the system and also locate which component has a fault. The second step is to adapt the controller to the faulty case so that the system is able to achieve its target. A linear matrix inequality (LMI) based feasible FTC method is applied such that the fault can be detected and diagnosed. An illustrated example is included to demonstrate the efficiency of the proposed algorithm, and satisfactory results have been obtained.     

一类纯反馈非仿射非线性系统的自适应神经网络变结构控制

研究了一类非仿射的纯反馈单输入单输出非线性系统。针对此系统,在中值定理、神经网络参数化和解耦Backstepping的基础上,提出了一种自适应变结构神经网络控制策略,而且所给出的定理证明闭环系统的所有信号在平衡点上是半全局一致有界的。通过对一个非仿射CSTR对象的仿真验证了该方法的有效性。     

基于自适应SSUKF的组合导航信息融合方法

针对车载组合导航系统噪声统计特性无法事先实时获取的问题,提出了一种神经网络辅助的自适应SSUKF信息融合算法.该算法利用神经网络在线估计系统噪声,采用SSUKF同时估计系统状态和在线训练神经网络的权值,从而能在系统噪声统计特性未知的情况下获得组合导航系统的实时最优估计,给出了算法的详细实现过程.最后,针对车载INS/GPS组合导航系统的信息融合问题进行了仿真研究.仿真结果表明,该算法在系统噪声统计特性未知的情况下仍能获得高精度的估计效果,同时与自适应UKF算法相比,有效降低了算法的计算量,提高了算法运行的实时性,证明了该算法是一种有效而实用的方法.     

Robust adaptive neural network control with supervisory controller

The problem of direct adaptive neural network control for a class of uncertain nonlinear systems with unknown constant control gain is studied in this paper. Based on the supervisory control strategy and the approximation capability of multilayer neural networks (MNNs), a novel design scheme of direct adaptive neural network controller is proposed.The adaptive law of the adjustable parameter vector and the matrix of weights in the neural networks and the gain of sliding mode control term to adaptively compensate for the residual and the approximation error of MNNs is determined by using a Lyapunov method. The approach does not require the optimal approximation error to be square-integrable or the supremum of the optimal approximation error to be known. By theoretical analysis, the closed-loop control system is proven to be globally stable in the sense that all signals involved are bounded, with tracking error converging to zero.Simulation results demonstrate the effectiveness of the approach.     

高阶非线性时滞系统的神经网络动态面控制

针对一类具有未知时变时滞的不确定高阶非线性系统,基于增加幂次积分方法,提出了一种非光滑状态反馈自适应神经网络动态面控制设计方案。通过构造适当的Lyapunov Krasovskii泛函处理了未知时变时滞不确定项;通过利用神经网络权值范数的适当形式幂次函数,将神经网络用于对在单步递推中所构造的未知函数进行建模;采用动态面技术,解决了“微分爆炸”问题。所提控制方案能够保证闭环控制系统的状态量和跟踪误差半全局一致终结有界。最后,仿真算例结果表明了该方案的有效性。     

发酵过程混合神经网络模型及其仿真

提出了一种新型的发酵过程混合神经网络模型，该模型由非线性神经网络和线性神经网络两部分组成，由于非线性神经网络采用结构具有线形式的Flat网络，两个网络能够合并为同一表达式，并具有线性形式，可采用线性最小二乘法求解网络权值，与串联结构及串并联结构混合神经网络模型相比，该模型训练方式简单，并可方便地使用在线辨识算法。     

协同随机微粒群优化的神经网络预测建模

针对一类难以精确建立数学模型的非线性控制系统,提出了协同随机微粒群优化CSPSO的神经网络预测建模方法.CSPSO在协同微粒群算法CPSO执行之后引入随机微粒群优化SPSO的思想,促使CPSO摆脱了伪最小值现象,并且保证其以概率1收敛于全局最优值.通过采集对象输入/输出数据,将CSPSO应用到模型权值的离线训练中,并给出了实现的具体步骤.结果表明在实验的几种算法中,CSPSO训练的神经网络模型精度较高且算法学习的稳定性较佳.     

非线性离散系统的直接自适应神经网络控制器

对于一类非仿射离散时间系统,提出了一种新的自适应神经网络控制器。首先推导与原系统等价的仿射形式模型,由仿射模型推导控制律。控制律中采用一个神经网络,与传统的基于反馈线性化的自适应神经网络设计方法中采用两个神经网络相比,计算量大大减少且避免了控制器奇异问题。神经网络权值根据系统输入输出信号进行更新,另外σ项的引入,取消了为保证参数收敛持续激励的条件。系统的稳定性通过Lyapunov方法进行了分析,仿真实例验证了控制器的有效性。     

一类输入受限的不确定非线性系统自适应Backstepping变结构控制

针对一类输入受限的不确定非线性系统,提出了一种自适应Backstepping变结构控制器设计方法。建立了受未知非线性特征约束的执行器故障模型,可以描述系统存在死区、齿隙、饱和、滞回等输入受限情形以及可能发生的执行器失效、卡死等故障情形。设计径向基函数神经网络补偿未建模动态项,引入一阶低通滤波器避免了Backstepping控制中的计算复杂性问题。自适应近似变结构控制能够有效削弱控制信号抖振。理论分析和仿真实验结果证明,提出的自适应鲁棒控制律能够在输入受限的情况下自适应地调节控制输入,使得闭环系统稳定且满足控制性能要求。     

一类非线性不确定系统自适应terminal滑模控制方法

针对一类非线性不确定系统设计了自适应terminal滑模控制器,使跟踪误差在有限时间内收敛到零,消除了通常滑模变结构控制的到达过程,因而闭环系统从t=0时刻就对干扰具有鲁棒性。采用RBF神经网络逼近系统未知的非线性函数,引入滑模误差对其权值进行在线自适应调整,改善动态性能。最后给出的仿真例子证明了算法的有效性。     

加工时间为连续随机变量的Job Shop调度问题

针对实际调度问题中存在的不确定现象,提出了加工时间服从正态分布、最大完成时间的期望值作为目标函数的随机Job Shop问题;然后提出了解决该问题的智能优化算法:采用随机模拟的方式产生输入输出数据,利用遗传算法训练神经网络,将训练过的神经网络嵌入到另一遗传算法中,用该遗传算法来优化Job Shop调度问题;最后给出了仿真实验,通过仿真实验证明,该算法对于解决加工时间为随机变量的Job Shop调度问题是行之有效的。     

基于干扰补偿的拦截弹新型反演姿态控制

针对同时具有未知干扰以及输入饱和与死区特性的大气层内拦截弹姿态控制系统,提出了一种基于干扰补偿的自适应动态面控制器设计方法。该方法通过设计改进的非线性干扰观测器(nonlinear disturbance observer, NDO)对未知干扰进行抑制,利用径向基函数(radial basis function,RBF)神经网络逼近输入饱和引起的非线性项,通过设计参数自适应律在线估计未知死区边界。通过构造合适的Lyapunov函数,证明闭环系统状态一致终结有界。仿真结果表明,所提方法鲁棒性良好,在输入非线性和未知干扰作用下,依然能良好地跟踪指令信号。     

Adaptive control of system with hysteresis using neural networks

1.INTRODUCTIONThe piezoelectric actuators are well suited for micro-position devices in precision engineering because oftheir fast response,nanometer resolution and biggerdriving force[1].However,hysteresis inherent topiezoelectric actuator severely li mits system’s perfor-mance such as giving rise to undesirable accuracy oroscillations,even leading to instability.Hysteresischaracteristics are generally nondifferentiable andusually unknown.It is a difficult task to mitigate itsharmful ef…     

Adaptive neural network tracking control for a class of unknown nonlinear time-delay systems

1 .INTRODUCTIONNeural network (NN) control has made great pro-gress in past decades[1 ~4]. In Ref .[1] , adaptivebounding design technique was applied to adaptiveneural control for a class of strict-feedback nonlin-ear systems . The requirement of a known boundon the network reconstruction error was removed.By introducing an integral Lyapunov function,anadaptive NN control approach was proposed forunknown strict-feedback nonlinear systems[2],where the controller singularity problem was …