共查询到19条相似文献,搜索用时 250 毫秒
1.
从改变仿射投影算法的条件出发,迫使仿射投影算法(APA)后验误差等于噪声,提出一种新的变阶仿射投影算法(E-APA)。该算法令迭代过程中阶数正比于误差向量的L2范数的平方与噪声功率之比,通过调节正则化因子来改变二者比值,从而实现了在初始阶段输入阶数较大,在收敛阶段输入阶数较小。实验结果表明该算法收敛速度快,稳态失调小,计算量少,优于目前最好的变阶算法(E-APA)。 相似文献
2.
迭代学习控制能够实现期望轨迹的完全跟踪而被广泛关注,但是采样迭代学习控制成果目前还比较少。针对一类有相对阶和输出延迟的非线性采样系统,研究了高阶迭代学习控制算法。利用Newton-Leibniz公式、贝尔曼引理和Lipschiz条件证明了当系统的采样周期足够小,迭代学习初态严格重复,且学习增益满足要求的条件,那么系统输出在采样点上收敛于期望输出。对一阶和二阶学习算法的仿真表明高阶算法在收敛速度上比一阶有明显改善。 相似文献
3.
Q(λ)学习算法是一种结合值迭代与随机逼近的思想的基于模型无关的多步离策略强化学习算法.针对经典的Q(λ)学习算法执行效率低、收敛速度慢的问题,从TD Error的角度出发,给出n阶TD Error的概念,并将n阶TD Error用于经典的Q(λ)学习算法,提出一种二阶TD Error快速Q(λ)学习算法——SOE-FQ(λ)算法.该算法利用二阶TD Error修正Q值函数,并通过资格迹将TD Error传播至整个状态动作空间,加快算法的收敛速度.在此基础之上,分析算法的收敛性及收敛效率,在仅考虑一步更新的情况下,算法所要执行的迭代次数T主要指数依赖于1/1-γ、1/ε.将SOE-FQ(λ)算法用于Random Walk和Mountain Car问题,实验结果表明,算法具有较快的收敛速度和较好的收敛精度. 相似文献
4.
时滞非线性系统的采样迭代学习控制 总被引:1,自引:0,他引:1
针对一类输入时滞非线性系统, 提出了一种采样迭代学习控制算法, 该算法不含跟踪误差的微分信号, 给出了学习算法收敛的充分条件, 当不存在初始误差、不确定扰动时, 算法在采样点处能实现对期望输出信号的完全跟踪, 否则, 跟踪误差一致有界, 仿真结果表明了该算法的有效性. 相似文献
5.
6.
针对一类非线性带扰动系统提出了高阶PID采样速代学习控制算法,讨论了高阶算法的收敛性问题以及该算法的优势与缺陷.与传统的证明方法不同,利用泰勒级数展开法证明了被控对象在输入干扰和输出测量噪声均有界的情况下,高阶PID采样速代学习控制算法的收敛性,并且得出了收敛条件.由于收敛条件中没有积分项,因此更加利于分析计算.与传统的一阶采样迭代学习控制算法相比,高阶采样迭代学习控制算法由于利用了更多先前的控制信息而能使被控对象的实际输出更加接近理想输出.给出了相应的数值仿真,证明了理论分析的有效性.与此同时,结合啤酒生产过程中糖化阶段中酒花添加等实际问题对该算法的应用前景作了一定的分析. 相似文献
7.
主要研究两相图像分割凸模型的三类快速数值算法.首先,分别针对无约束和有约束的图像分割凸模型分别提出相应的具有O(1/k)阶收敛速率的梯度投影算法,并结合快速迭代收缩算法的加速收敛策略,将所提出的梯度投影算法的收敛速率从O(1/k)阶提高到O(1/k2)阶;其次,基于分块协调下降的思想,对无约束的图像分割凸模型采用Newton法求解,该算法不仅是单调下降的,而且具有二阶收敛性;然后,根据交互式迭代算法的思想,在约束模型的Fenchel原始-对偶形式的基础上,提出了一种通过原始变量和对偶变量交互式混合迭代求解的算法,所提出的算法在求解过程中避免了梯度算子和散度算子作用于未知变量,使得迭代形式更简单;最后,仿真实验表明了这3类算法的有效性和在收敛速率上的优势. 相似文献
8.
9.
算法的迭代步长对于算法的收敛性能有着重要影响。针对固定步长的非线性主成分分析(NPCA)算法不能兼顾收敛速度和估计精度的情形,提出基于梯度的自适应变步长NPCA算法和最优变步长NPCA算法两种自适应变步长算法来改善其收敛性能。特别地,最优变步长NPCA算法通过对代价函数进行一阶线性近似表示,从而计算出当前的最优迭代步长。该算法的迭代步长随估计误差的变化而变化,估计误差大,迭代步长相应大,反之亦然;且不需要人工设置任何参数。仿真结果表明,当算法的估计精度相同时,与固定步长NPCA算法相比,两种自适应变步长NPCA算法相对固定步长NPCA算法都具有更好的收敛速度或跟踪性能,且最优变步长NPCA算法的性能优于基于梯度的自适应变步长NPCA算法。 相似文献
10.
具有扰动的非线性系统高阶迭代学习控制 总被引:1,自引:0,他引:1
李宏胜 《模式识别与人工智能》2006,19(4)
迭代学习控制(ILC)利用系统的重复性不断改进控制性能.本文讨论一类具有扰动的非线性、时变系统高阶迭代学习控制算法及其迭代学习收敛的充分条件,并与D型迭代学习算法相比,讨论典型PD高阶ILC算法的收敛速度.仿真结果证实高阶ILC算法具有更快的收敛速度,并且当系统满足收敛条件、不确定项及输出扰动项有界时迭代学习收敛. 相似文献
11.
Ying-Ying Liu 《International journal of systems science》2013,44(9):1728-1740
In this article, the problem of sampled-data H ∞ control for networked control systems (NCSs) with digital control inputs is considered, where the physical plant is modelled as a continuous-time one, and the control inputs are discrete-time signals. By exploiting a novel Lyapunov–Krasovskii functional, using the Leibniz–Newton formula and a free-weighting matrix method, sufficient conditions for sampled-data H ∞ performance analysis and H ∞ controller design for such systems are given. Since the obtained conditions of H ∞ controller design are not expressed strictly in term of linear matrix inequalities, the sampled-data H ∞ controller is solved using modified cone complementary linearisation algorithm. In addition, the new sampled-data stability criteria for the NCSs is proved to be less conservative than some existing results. Numerical examples demonstrate the effectiveness of the proposed methods. 相似文献
12.
研究一般非均匀采样数据系统鲁棒传感器故障检测设计问题.首先,基于输出时滞方法将非均匀采样数据系统转换成具有时变时滞输出的连续系统;然后,选择故障检测滤波器作为残差产生器,并将故障检测设计问题描述成一个多目标优化问题,即连续时间过程噪声和离散时间测量噪声对残差信号的H∞范数小于一个给定值,同时传感器故障对残差信号的l2增益大于一个给定值,基于输入输出方法以矩阵不等式的形式给出该多目标优化问题有解的充分条件;进一步的,提出一个迭代算法来权衡噪声鲁棒性与故障灵敏度,并将矩阵不等式转换成可解的线性矩阵不等式.最后,对某型飞控系统的仿真实验验证了所提方法的有效性. 相似文献
13.
In this paper, we extend the deterministic learning theory to sampled-data nonlinear systems. Based on the Euler approximate model, the adaptive neural network identifier with a normalized learning algorithm is proposed. It is proven that by properly setting the sampling period, the overall system can be guaranteed to be stable and partial neural network weights can exponentially converge to their optimal values under the satisfaction of the partial persistent excitation (PE) condition. Consequently, locally accurate learning of the nonlinear dynamics can be achieved, and the knowledge can be represented by using constant-weight neural networks. Furthermore, we present a performance analysis for the learning algorithm by developing explicit bounds on the learning rate and accuracy. Several factors that influence learning, including the PE level, the learning gain, and the sampling period, are investigated. Simulation studies are included to demonstrate the effectiveness of the approach. 相似文献
14.
15.
This paper addresses the problems of the input-to-state practical stability (ISPS) analysis and output feedback controller design for switched affine systems (SASs) subject to external disturbances. First, a switched affine observer is developed to estimate unmeasurable states. Then by combining the sampled-data control approach, a less conservative mode-dependent dynamic event-triggered mechanism (ETM) is established. The proposed dynamic ETM cannot only avoid Zeno behavior but also reduce the network transmission burden effectively. Further, based on time-dependent Lyapunov-Krasovskii functional and state-dependent switching laws, a set of feasible ISPS conditions are presented in the LMI forms by means of singular value decomposition. The designed switching law depends upon the sampled-data information of the estimated state and gets rid of the chattering phenomenon. Finally, an application example of the DC-DC flyback converter is given to verify the efficacy of the proposed algorithm. 相似文献
16.
Networked Iterative Learning Control Design for Nonlinear Systems with Stochastic Output Packet Dropouts
下载免费PDF全文
![点击此处可从《Asian journal of control》网站下载免费的PDF全文](/ch/ext_images/free.gif)
This paper develops two proportional‐type (P‐type) networked iterative learning control (NILC) schemes for a class of discrete‐time nonlinear systems whose stochastic output packet dropouts are modeled as 0‐1 Bernoulli stochastic sequences. In constructing the NILC schemes, two kinds of compensation algorithm of the dropped outputs are given. One is to replace the instant‐wise dropped output data with the synchronous desired output data; the other is to substitute the dropped data with the consensus‐instant output data used at the previous iteration. By adopting the lifting technique, it is derived that under certain conditions the expectations of the tracking errors incurred by the proposed NILC schemes converge to zero along the iteration axis. Numerical experiments are carried out for validity and effectiveness. 相似文献
17.
对于非线性迭代学习控制问题,提出基于延拓法和修正Newton法的具有全局收敛性的迭代学习控制新方法.由于一般的Newton型迭代学习控制律都是局部收敛的,在实际应用中有很大局限性.为拓宽收敛范围,该方法将延拓法引入迭代学习控制问题,提出基于同伦延拓的新的Newton型迭代学习控制律,使得初始控制可以较为任意的选择.新的迭代学习控制算法将求解过程分成N个子问题,每个子问题由换列修正Newton法利用简单的递推公式解出.本文给出算法收敛的充分条件,证明了算法的全局收敛性.该算法对于非线性系统迭代学习控制具有全局收敛和计算简单的优点. 相似文献
18.
This paper proposes two kinds of iterative learning control (ILC) schemes for a class of the distributed parameter systems based on sensor–actuator networks which can be described by hyperbolic partial differential equations. A D-type ILC algorithm is first considered and the convergent condition of the output error is obtained via the contraction mapping methodology. Then, the PD-type ILC algorithm is considered in this hyperbolic distributed parameter systems based on sensor–actuator networks. Finally, a cable equation with air and structural damping is given to illustrate the effectiveness of the proposed methods. 相似文献
19.
Jun Yoneyama 《Applied Soft Computing》2011,11(1):249-255
In practice, the system is often modeled as a continuous-time fuzzy system, while the control input is applied only at discrete instants. This system is called a sampled-data control system. In this paper, robust guaranteed cost control for uncertain sampled-data fuzzy systems is discussed. A guaranteed cost control where a quadratic cost function is bounded by a certain scalar, not only stabilizes a system but also considers a control performance. A typical sampled-data control is the zero-order input, which can be represented as a piecewise-continuous delay. Here we take a delay system approach to the sampled-data guaranteed cost control problem. The closed-loop system with a sampled-data state feedback controller becomes a system with time-varying delay. First, guaranteed cost control performance conditions for the closed-loop system are given in terms of linear matrix inequalities (LMIs). Such conditions are derived by using Leibniz–Newton formula and free weighting matrix method for fuzzy systems under the assumption that sampling time is not greater than some prescribed scalar. Then, a design method of robust guaranteed cost state feedback controller for uncertain sampled-data fuzzy systems is proposed. Examples are given to illustrate our robust sampled-data guaranteed cost control design. 相似文献