基于神经网络的模型跟随鲁棒自适应控制

针对一类复杂非线性动力学系统,提出一种基于神经网络动态补偿的模型跟随非线 性鲁棒自适应控制策略.采用神经网络在线补偿控制器以克服系统的未建模动力学和非线性 耦合因素的影响,从而提高了模型跟随控制的动态性能和稳态精度;当系统存在模型不确定 性和外部扰动时,其输出仍能精确地跟踪期望参考模型的输出.同时给出了闭环误差系统鲁 棒稳定性的证明.应用示例表明,所提方法可保证闭环系统具有良好的跟踪性能和鲁棒性,且 算法简单,易于在线控制.     

一类非线性系统的自组织模糊CMAC神经网络自适应重构跟踪控制

提出了一种非线性系统的自组织模糊CMAC(SOFCMAC)神经网络自适应重构跟踪控制方法,首先通过构造增广系统,设计出线性渐近跟踪控制器,然后采用SOFCMAC神经网络在线重构系统的非线性特性,以消除非线性特性引起的系统误差,可保证非线性系统闭环稳定并使系统输出跟踪期望输出.仿真算例证明了SOFCMAC神经网络自适应重构跟踪控制系统的稳定性.     

基于RBF神经网络的电动负载模拟器摩擦与间隙补偿方法研究

为提高电动负载模拟器系统的动态性能和信号跟踪准确度,提出针对系统摩擦和间隙进行补偿的方法。采用基RBF波神经网络的PID 控制器实现摩擦非线性补偿,同时利用间隙逆模型针对间隙非线性进行补偿。利用Matlab 软件对补偿结果进行仿真验证,仿真结果显示经过补偿后系统正弦响应曲线跟随性能变好,跟踪误差明显减小,准确度得到很大改善。仿真结果证明：基于RBF神经网络的PID 控制器和间隙逆模型分别对摩擦和间隙有明显的抑制效果,系统动态性能得到提高。     

基于逆模型辨识的Wiener型传感器动态补偿研究

在测量系统中许多传感器动态特性是一个非线性Wiener模型,即存在着严重的静态非线性和动态响应滞后.为了补偿动态误差,采用模型参考和Wiener逆模型辨识的算法建立动态补偿单元.补偿单元由一个静态逆模型和动态逆模型构成.通过静态标定方法,采用单输入/单输出的模糊小脑神经网络(SISO-FCMAC)建立传感器静态非线性模...     

基于BP神经网络的自适应补偿控制方法

研究工业过程控制系统补偿问题,对于一类模型未知的SISO非线性系统,传统的控制方法不能获得被控系统的精确数学模型,因而在系统稳定性和鲁棒性上存在缺馅,控制效果不佳。为了提高被控非线性系统的稳定性和鲁棒性,提出了一种基于BP神经网络的自适应补偿控制方法。首先,通过逆系统理推导了被控系统输出和伪控制量之间的误差,然后误差进行在线自适应BP神经网络补偿,从而实现对被控系统的BP神经网络自适应补偿控制,且采用Lyapunov理论证明BP神经而网络的收敛性和闭环系统的稳定性。计算机仿真表明所提方法明显提高了非线性系统的鲁棒控制性能。     

严格反馈非线性系统的自适应神经网络输出反馈控制

基于非线性反馈函数,文章设计神经网络状态观测器,解决一类非线性系统的输出反馈控制问题.非线性反馈神经网络观测器在系统存在不确定性函数的情况下实时估计系统状态.利用所获得的状态信号,设计了自适应神经网络动态面控制器,同时保证了闭环系统的稳定性和所有信号的有界性.通过调节设计参数的取值能够达到期望的闭环跟踪性能.数值仿真表明,所设计的状态观测器不需要对原系统做状态变换,能够克服输出反馈滑模控制器带来的抖震问题.     

输入死区下的多输入多输出系统自适应神经网络容错控制

针对一类具有传感器故障和不对称输入死区的非线性多输入多输出非严格反馈系统,本文提出一种自适应神经网络容错控制方案.控制器的设计以反步法为框架,采用自适应神经网络控制方法处理传感器故障,利用死区斜率的有界性补偿输入死区对系统性能造成的影响,同时引入动态面控制技术克服“计算爆炸”的问题.该控制方法不仅能够保证闭环系统中所有...     

基于状态全反馈的动态迟滞非线性系统

该文针对不平滑、多映射动态迟滞非线性系统,提出了一种基于神经网络自适应控制方案.在该方案中,通过利用神经网络来逼近模型误差,避免了目前常用逆模型补偿方案中,需求取复杂逆模型的问题.应用Lyapnov稳定定理,证明了整个闭环系统的跟踪误差及神经网络权值将收敛到零点一个有界邻域内.仿真结果表明,所提出的控制方案能够有效补偿迟滞非线性对系统的影响.     

基于UKF 和神经网络的一类非线性系统状态估计

在模型未知的情况下,估计过程的重要变量尤为重要.鉴于此,采用不敏卡尔曼滤波(UKF)与神经网络相结合的方法,解决一类未知模型非线性系统的状态估计问题.采用动态神经网络对非线性系统进行建模,利用UKF对状态和权值进行同时更新,从而达到神经网络逼近真实模型,估计值跟随真实值的目的.通过两个仿真实例表明了所提出的方法具有良好的估计效果,并且状态在输出中的比重越大,其估计精度越高.     

基于预估器的一类多智能体系统神经动态面输出一致控制

本文针对一类含未知扰动与非对称输入饱和的非线性多智能体系统,提出基于预估器的神经动态面输出一致控制策略.在设计预估器的基础上构造预估误差,驱动神经网络更新权值估计系统未知动态,并将预估器与神经网络应用于非线性扰动观测器来补偿广义扰动.本文所提出的控制策略采用神经网络权值范数学习方法,减少学习参数数目.对于非对称的输入饱和,设计辅助系统,其生成的辅助变量与反步法相结合补偿输入限制.结合图论知识和Lyapunov函数等技术,证明多智能体系统的输出一致跟踪误差以及闭环系统中的所有信号最终有界.最后通过一组四旋翼飞行器和数值仿真验证提出控制策略的有效性.     

A RBF neural network sliding mode controller for SMA actuator

A radial basis function neural network sliding-mode controller (RBFSMC) is proposed to control a shape memory alloy (SMA) actuator. This approach, which combines a RBF neural network with sliding-mode control (SMC), is presented for the tracking control of a class of nonlinear systems having parameter uncertainties. The centers and output weights of the RBF neural network are updated through on-line learning, which causes the output of the neural network control to approximate the sliding-mode equivalent control along the direction that makes the sliding-mode asymptotically stable. Using Lyapunov theory, the asymptotic stability of the overall system is proven. Then, the controller is applied to compensate for the hysteresis phenomenon seen in SMA. The results show that the controller was applied successfully. The control results are also compared to those of a conventional SMC.     

基于神经网络的不确定机器人自适应滑模控制

提出一种机器人轨迹跟踪的自适应神经滑模控制。该控制方案将神经网络的非线性映射能力与变结构控制理论相结合,利用RBF网络自适应学习系统不确定性的未知上界,神经网络的输出用于自适应修正控制律的切换增益。这种新型控制器能保证机械手位置和速度跟踪误差渐近收敛于零。仿真结果表明了该方案的有效性。     

一类不确定非线性系统的鲁棒自适应轨迹线性化控制

针对一类不确定非线性系统,研究了一种新的鲁棒自适应轨迹线性化控制方案.利用径向基神经网络的在线逼近能力以及被控对象分析模型的有用信息设计一种径向基神经网络干扰观测器来估计系统中存在的不确定性.观测器输出用于设计补偿控制律抵消不确定性对系统性能的影响,鲁棒自适应控制律用于克服逼近误差.采用Lyapunov方法严格证明了在自适应调节律作用下闭环系统所有误差信号最终有界.最后利用倒立摆系统验证了新方法的有效性.     

基于自适应神经网络的不确定非线性系统的模糊跟踪控制

提出了一种基于模糊模型和自适应神经网络的跟踪控制方法.在系统具有未知不确定非线性特性的情况下,首先利用T＿S模糊模型对系统的已知特性进行近似建模,对基于模糊模型的模糊H∞跟踪控制律进行输出跟踪控制.并在此基础上,进一步采用RBF神经网络完全自适应控制,通过在线自适应调整RBF神经网络的权重、函数中心和宽度,从而有效地消除系统的未知不确定性和模糊建模误差的影响,保证了非线性闭环系统的稳定性和系统的H∞跟踪性能,而不要求系统的不确定项和模糊建模误差满足任何匹配条件或约束.最后,将所提出的方法应用到一非线性混沌系统,仿真结果表明了所提出的方案不仅能够有效地稳定该混沌系统,而且能使系统输出跟踪期望输出.     

Filtering adaptive neural network controller for multivariable nonlinear systems with mismatched uncertainties

This paper synthesizes a filtering adaptive neural network controller for multivariable nonlinear systems with mismatched uncertainties. The multivariable nonlinear systems under consideration have both matched and mismatched uncertainties, which satisfy the semiglobal Lipschitz condition. The nonlinear uncertainties are approximated by a Gaussian radial basis function (GRBF)‐based neural network incorporated with a piecewise constant adaptive law, where the adaptive law will generate adaptive parameters by solving the error dynamics between the real system and the state predictor with the neglection of unknowns. The combination of GRBF‐based neural network and piecewise constant adaptive law relaxes hardware limitations (CPU). A filtering control law is designed to handle the nonlinear uncertainties and deliver a good tracking performance with guaranteed robustness. The matched uncertainties are cancelled directly by adopting their opposite in the control signal, whereas a dynamic inversion of the system is required to eliminate the effect of the mismatched uncertainties on the output. Since the virtual reference system defines the best performance that can be achieved by the closed‐loop system, the uniform performance bounds are derived for the states and control signals via comparison. To validate the theoretical findings, comparisons between the model reference adaptive control method and the proposed filtering adaptive neural network control architecture with the implementation of different sampling time are carried out.     

机器人的非奇异终端神经滑模轨迹跟踪控制

针对具有不确定性的机器人系统,为提高系统的稳态跟踪精度,提出一种非奇异终端神经滑模轨迹跟踪控制方案.控制器采用改进的非奇异终端滑模面,并基于径向基函数神经网络自适应调整控制律的切换项,不但克服了在设计中需要知道系统不确定性的上界的限制,而且平滑了控制信号.可应用Lyapunov稳定性理论证明了系统的渐近稳定性和跟踪误差的渐近收敛性.仿真结果验证了控制方法不仅能够保证机器人系统轨迹跟踪控制的快速性和鲁棒性,而且有效地削弱了抖振,可见方案是可行且有效的.     

Direct adaptive neural control of nonlinear systems with extreme learning machine

A direct adaptive neural control scheme for a class of nonlinear systems is presented in the paper. The proposed control scheme incorporates a neural controller and a sliding mode controller. The neural controller is constructed based on the approximation capability of the single-hidden layer feedforward network (SLFN). The sliding mode controller is built to compensate for the modeling error of SLFN and system uncertainties. In the designed neural controller, its hidden node parameters are modified using the recently proposed neural algorithm named extreme learning machine (ELM), where they are assigned random values. However, different from the original ELM algorithm, the output weight is updated based on the Lyapunov synthesis approach to guarantee the stability of the overall control system. The proposed adaptive neural controller is finally applied to control the inverted pendulum system with two different reference trajectories. The simulation results demonstrate good tracking performance of the proposed control scheme.     

一类非线性不确定系统的神经网络控制

针对一类非线性不确定系统,提出了一种自适 应神经网络控制方案．被控系统是部分已知的,其中系统已知的动态特性被用来设计保证标 称模型稳定的反馈控制器,而基于神经网络的动态补偿器则用于补偿系统的非线性不确定性 ,从而可以保证系统输出跟踪误差渐近收敛于0．     

Decentralized adaptive neural network control of cascaded DC–DC converters with high voltage conversion ratio

Decentralized output voltage tracking of cascaded DC–DC converters is an interesting topic to obtain a high voltage conversion ratio. The control purpose is challenging due to the load resistance changes, renewable energy supply voltage variations and interaction of the individual converters. In this paper, four novel decentralized adaptive neural network controllers are designed on the cascaded DC–DC buck and boost converters under load and DC supply voltage uncertainties. In the beginning, individual buck and boost converter average models that can operate in both continuous and discontinuous conduction modes are derived. Then, the interconnected and decentralized state-space models of cascaded buck and boost converters are extracted. These models are highly nonlinear with unknown uncertainties which can be estimated by neural networks. Further, two decentralized adaptive backstepping neural network voltage controllers are proposed on cascaded buck converters to deal with uncertainties and interactions. However, these control strategies are not applicable to a boost converter due to its non-minimum phase nature. Then, two novel decentralized adaptive neural network with a conventional proportional–integral reference current generator are developed on the cascaded boost converters. Practical stability of the overall system is guaranteed for the proposed controllers using Lyapunov stability theorem. Finally, four control strategies provide good quality of output voltage in the presence of uncertainties and interactions. Comparative simulations are carried out on cascaded buck and boost converters to validate the effectiveness and performance of the designed methods.     

Robust adaptive fault accommodation for a robot system using a radial basis function neural network

A discrete-time radial basis function (RBF) neural network is designed for the fault accommodation of robotic systems. A robust learning algorithm using the adaptive dead-zone technique is presented to train the network parameters (weights and centres). This scheme assures the convergence of the estimate errors of both the neural network and the fault-monitoring system in the presence of system uncertainties. Simulations have been done on applying the RBF-network-based fault accommodation scheme to a two-link robotic manipulator. The main advantage of the adaptive algorithm is that the upper bound of system uncertainties is not known in advance, which makes the system more practical for the fault accommodation scheme as demonstrated.