对称工件定位算法:收敛性及其改进

Symmetric workpiece localization algorithms combine alternating optimization and linearization. The iterative variables are partitioned into two groups. Then simple optimization approaches can be employed for each subset of variables, where optimization of configuration variables is simplified as a linear least-squares problem (LSP). Convergence of current symmetric localization algorithms is discussed firstly. It is shown that simply taking the solution of the LSP as start of the next iteration may result in divergence or incorrect convergence. Therefore in our enhanced algorithms, line search is performed along the solution of the LSP in order to find a better point reducing the value of objective function. We choose this point as start of the next iteration. Better convergence is verified by numerical simulation. Besides, imposing boundary constraints on the LSP proves to be another efficient way.     

Robust H-infinity control of uncertain stochastic time-delay linear repetitive processes

Repetitive processes are a distinct class of 2D systems of both theoretic and practical interest. The robust H-infinity control problem for uncertain stochastic time-delay linear continuous repetitive processes is investigated in this paper. First, sufficient conditions are proposed in terms of stochastic Lyapunov stability theory, Itˆo differential rule and linear matrix inequality technology. The corresponding controller design is then cast into a convex optimization problem. Attention is focused on constructing an admissible controller, which guarantees that the closed-loop repetitive processes are mean-square asymptotically stable and have a prespecified H-infinity performance γ with respect to all energy-bounded input signals. A numerical example illustrates the effectiveness of the proposed design scheme.     

线性离散奇异系统的不定二次最优控制问题: Krein空间方法

The finite time horizon indefinite linear quadratic(LQ) optimal control problem for singular linear discrete time-varying systems is discussed. Indefinite LQ optimal control problem for singular systems can be transformed to that for standard state-space systems under a reasonable assumption. It is shown that the indefinite LQ optimal control problem is dual to that of projection for backward stochastic systems. Thus, the optimal LQ controller can be obtained by computing the gain matrices of Kalman filter. Necessary and sufficient conditions guaranteeing a unique solution for the indefinite LQ problem are given. An explicit solution for the problem is obtained in terms of the solution of Riccati difference equations.     

Lifted System Framework for Learning Control with Different Trial Lengths

This paper addresses an iterative learning control(ILC) design problem for discrete-time linear systems with randomly varying trial lengths. Due to the variation of the trial lengths, a stochastic matrix and an iteration-average operator are introduced to present a unified expression of ILC scheme. By using the framework of lifted system, the learning convergence condition of ILC in mathematical expectation is derived without using λ-norm. It is shown that the requirement on classic ILC that all trial lengths must be identical is mitigated and the identical initialization condition can be also removed. In the end, two illustrative examples are presented to demonstrate the performance and the effectiveness of the proposed ILC scheme for both time-invariant and time-varying linear systems.     

An Adaptive Learning Method for the Generation of Fuzzy Inference System from Data

Designing a fuzzy inference system （FIS） from data can be divided into two main phases： structure identification and parameter optimization. First, starting from a simple initial topology, the membership functions and system rules are defined as specific structures. Second, to speed up the convergence of the learning algorithm and lighten the oscillation, an improved descent method for FIS generation is developed. Furthermore, the convergence and the oscillation of the algorithm are systematically analyzed. Third, using the information obtained from the previous phase, it can be decided in which region of the input space the density of fuzzy rules should be enhanced and for which variable the number of fuzzy sets that used to partition the domain must be increased. Consequently, this produces a new and more appropriate structure. Finally, the proposed method is applied to the problem of nonlinear function approximation.     

Delay-dependent robust stabilization and H ∞ control for uncertain stochastic T-S fuzzy systems with discrete interval and distributed time-varying delays

In this paper, delay-dependent robust stabilization and H∞ control for uncertain stochastic Takagi-Sugeno (T-S) fuzzy systems with discrete interval and distributed time-varying delays are discussed. The purpose of the robust stochastic stabilization problem is to design a memoryless state feedback controller such that the closed-loop system is mean-square asymptotically stable for all admissible uncertainties. In the robust H∞ control problem, in addition to the mean-square asymptotic stability requirement, a prescribed H∞ performance is required to be achieved. Sufficient conditions for the solvability of these problems are proposed in terms of a set of linear matrix inequalities (LMIs) and solving these LMIs, a desired controller can be obtained. Finally, two numerical examples are given to illustrate the effectiveness and less conservativeness of our results over the existing ones.     

具有适应性突变和惯性权重的粒子群优化(PSO)算法及其在动态系统参数估计中的应用

An important problem in engineering is the unknown parameters estimation in nonlinear systems. In this paper, a novel adaptive particle swarm optimization (APSO) method is proposed to solve this problem. This work considers two new aspects, namely an adaptive mutation mechanism and a dynamic inertia weight into the conventional particle swarm optimization (PSO) method. These mechanisms are employed to enhance global search ability and to increase accuracy. First, three well-known benchmark functions namely Griewank, Rosenbrock and Rastrigrin are utilized to test the ability of a search algorithm for identifying the global optimum. The performance of the proposed APSO is compared with advanced algorithms such as a nonlinearly decreasing weight PSO (NDWPSO) and a real-coded genetic algorithm (GA), in terms of parameter accuracy and convergence speed. It is confirmed that the proposed APSO is more successful than other aforementioned algorithms. Finally, the feasibility of this algorithm is demonstrated through estimating the parameters of two kinds of highly nonlinear systems as the case studies.     

Optimal linear estimators for systems with random measurement delays

This paper is concerned with the optimal linear estimation problem for linear discrete-time stochastic systems with random measurement delays. A new model that describes the random delays is constructed where possible the largest delay is bounded. Based on this new model, the optimal linear estimators including filter, predictor and smoother are developed via an innovation analysis approach. The estimators are recursively computed in terms of the solutions of a Riccati difference equation and a Lyapunov difference equation. The steady-state estimators are also investigated. A sufficient condition for the convergence of the optimal linear estimators is given. A simulation example shows the effectiveness of the proposed algorithms.     

A new approximation algorithm for sorting of signed permutations

Sequence comparison leads to a combinatorial optimization problem of sorting permutations by reversals and transpositions.Namely,given any two permutations,find the shortest distance between them.This problem is related with genome rearrangement,genes are oriented in DNA sequences.The transpositions which have been studied in the liteature can be viewed as operations working on two consecutive segments of the genome.In this paper,a new kind of transposition which can work on two arbitrary segments of the genome is proposed,and the sorting of signed permutations by reversals and this new kind of transpostitions are studied.After establishing a lower bound on the number of operations needed,a 2-approximation algorithm is presented for this problem and an example is given to show that the performance ratio of the algorithm cannot be improved.     

A Near-Optimal Optimization Algorithm for Link Assignment in Wireless Ad-Hoc Networks

Over the past few years, wireless networking technologies have made vast forays in our daily lives. In wireless ad-hoc networks, links are set up by a number of units without any permanent infrastructures. In this paper, the resource optimization is considered to maximize the network throughput by efficiently using the network capacity, where multi-hop functionality and spatial TDMA （STDMA） access scheme are used. The objective is to find the minimum frame length with given traffic distributions and corresponding routing information. Because of the complex structure of the underlying mathematical problem, previous work and analysis become intractable for networks of realistic sizes. The problem is addressed through mathematical programming approach, the linear integer formulation is developed for optimizing the network throughput, and then the similarity between the original problem and the graph edge coloring problem is shown through the conflict graph concept. A column generation solution is proposed and several enhancements are made in order to fasten its convergence. Numerical results demonstrate that the theoretical limit of the throughput can be efficiently computed for networks of realistic sizes.     

机器学习随机优化方法的个体收敛性研究综述

随机优化方法是求解大规模机器学习问题的主流方法,其研究的焦点问题是算法是否达到最优收敛速率与能否保证学习问题的结构。目前,正则化损失函数问题已得到了众多形式的随机优化算法,但绝大多数只是对迭代进行 平均的输出方式讨论了收敛速率,甚至无法保证最为典型的稀疏结构。与之不同的是,个体解能很好保持稀疏性,其最优收敛速率已经作为open问题被广泛探索。另外,随机优化普遍采用的梯度无偏假设往往不成立,加速方法收敛界中的偏差在有偏情形下会随迭代累积,从而无法应用。本文对一阶随机梯度方法的研究现状及存在的问题进行综述,其中包括个体收敛速率、梯度有偏情形以及非凸优化问题,并在此基础上指出了一些值得研究的问题。     

嵌入层叠混沌策略的随机粒子群算法

鉴于求解复杂问题时粒子群优化算法易出现早熟收敛的问题,通过引入轨迹扰动因子,提出随机粒子群进化迭代方程.该方程在统计行为中保证粒子向特定的收敛中心逼近,但对“旧址”的依赖性呈现出随机特性,从而使粒子群的快速跳转和迁移成为可能,避免过早落入局部陷阱.同时该进化方程还利用层叠混沌策略和对称极值扰动策略进一步增强算法的局部收敛性和全局搜索性.实验表明,由上述进化方程和改进策略构成的随机混沌粒子群算法具有鲁棒性较强、收敛速度较快和精度较高等优势,性能优于其他同源粒子群算法.     

Convergence and stability of distributed stochastic iterativeprocesses

A mathematical framework is proposed for convergence analysis of stochastic iterative processes arising in applications of pseudogradient optimization algorithms. The framework is based upon the concepts of stochastic difference inequalities and vector Lyapunov functions, which are ideally suited for reducing the dimensionality problem arising in testing convergence of distributed parallel schemes. By applying the M-matrix conditions to a test matrix having a dimension equal to the number of the processor in the scheme, one can use the framework to select suitable scaling factors for each individual processor, producing a satisfactory convergence rate of the overall iterative process. The results provide new convergence tests for distributed iterative processes arising in decentralized extremal regulation, adaptation, and parameter estimation schemes     

基于次梯度的L1正则化Hinge损失问题求解研究

Hinge损失函数是支持向量机(support vector machines,SVM)成功的关键,L1正则化在稀疏学习的研究中起关键作用.鉴于两者均是不可导函数,高阶梯度信息无法使用.利用随机次梯度方法系统研究L1正则化项的Hinge损失大规模数据问题求解.首先描述了直接次梯度方法和投影次梯度方法的随机算法形式,并对算法的收敛性和收敛速度进行了理论分析.大规模真实数据集上的实验表明,投影次梯度方法对于处理大规模稀疏数据具有更快的收敛速度和更好的稀疏性.实验进一步阐明了投影阈值对算法稀疏度的影响.     

A convergence analysis of unconstrained and bound constrained evolutionary pattern search

We present and analyze a class of evolutionary algorithms for unconstrained and bound constrained optimization on R(n): evolutionary pattern search algorithms (EPSAs). EPSAs adaptively modify the step size of the mutation operator in response to the success of previous optimization steps. The design of EPSAs is inspired by recent analyses of pattern search methods. We show that EPSAs can be cast as stochastic pattern search methods, and we use this observation to prove that EPSAs have a probabilistic, weak stationary point convergence theory. This convergence theory is distinguished by the fact that the analysis does not approximate the stochastic process of EPSAs, and hence it exactly characterizes their convergence properties.     

Robust real-time identification of linear systems with correlated noise

The problem of recursive robust identification of linear discrete-time single-input single-output dynamic systems with correlated disturbances is considered. Problems related to the construction of optimal robust stochastic approximation algorithms in the min-max sense are demonstrated. Since the optimal solution cannot be achieved in practice, several robustified stochastic approximation algorithms are derived on the basis of a suitable non-linear transformation of normalized residuals, as well as step-by-step optimization with respect to the weighting matrix of the algorithm. The convergence of the developed algorithms is established theoretically using the ordinary differential equation approach. Monte Carlo simulation results are presented for the quantitative performance evaluation of the proposed algorithms. The results indicate the most suitable algorithms for applications in engineering practice.     

多种群退火贪婪混合遗传算法

遗传算法是应用比较广泛的一种随机优化算法,遗传算法的收敛速度与问题解的质量是影响算法寻优性能的一对主要矛盾。为了提高遗传算法的性能,论文通过将局部搜索能力较强的贪婪算法引入遗传算法,并且同模拟退火和多种群并行遗传进化思想有机结合起来的方法,提出了一个改进型的算法——多种群退火贪婪混合遗传算法(MultigroupAnnealingGreedyHybridGeneticAlgorithm,简称MAGHGA)。仿真结果表明,该算法避免了在遗传算法中存在的早熟收敛问题,增强了算法的全局收敛性,同时也有效地提高了算法的收敛速度。     

离散萤火虫优化算法求解概率旅行商问题

对随机组合优化问题中的概率旅行商问题(PTSP)的理论和方法进行了研究分析,采用现代进化算法中有代表性发展优势的萤火虫优化算法(FA),提出一种离散萤火虫优化算法(DFA)以求解.其中引入了新的学习机制使其相比原始的萤火虫优化算法,更容易搜索到全局最优解,有更好的收敛性能.实验中用TSPLIB中的经典实例进行测试来验证其可行性.考察了萤火虫数量和进化迭代次数对求解结果性能的影响,并将DFA与GA、PSO和ACO等其他著名的进化计算算法进行性能比较.实验结果证实了DFA无论对固定访问概率,还是访问概率为区间内随机数等不同情况,都具有良好的有效性和高效性,因此对求解随机组合优化系列问题的有效解决具有一定参考和借鉴价值.     

基于Logistic模型和随机差分变异的正弦余弦算法

针对标准正弦余弦算法(Sine Cosine Algorithm,SCA)处理全局优化问题时存在收敛速度慢、易陷入局部最优和求解精度低的缺点,文中提出了一种基于非线性转换参数和随机差分变异策略的改进正弦余弦算法(LS-SCA)。首先,设计一种基于Logistic模型的非线性转换参数策略以平衡算法的全局搜索和局部开发能力;其次,引入随机差分变异策略以增强种群的多样性与避免算法陷入局部最优;最后,将非线性转换参数和随机差分变异策略进行融合。一方面,选取12个标准测试函数进行全局寻优的仿真实验。结果表明,与其他SCA类算法和最新智能算法相比,LS-SCA在收敛精度和收敛速度指标上均能达到较优的效果。其中,随机差分变异策略对LS-SCA全局寻优能力的提升尤为明显。另一方面,利用LS-SCA优化神经网络参数解决了两类经典分类问题。实验结果表明,与传统的BP算法和其他智能算法相比,基于LS-SCA的神经网络能达到较高的分类准确率。     

Optimization of control parameters in genetic algorithms: A stochastic approach

This paper proposes a stochastic approach for optimization of control parameters ( probabilities of crossover and mutation ) in genetic algorithms ( GAs ) . The genetic search can be modelled as a controlled Markovian process, the transition of which depends on the control parameters. A stochastic optimization problem is formed for control of GA parameters, based on a given performance index of populations and analysed as a controlled Markovian process during the genetic search. The optimal values of control parameters can be found from a recursive estimation of control parameters, which is obtained by introducing a stochastic gradient of the performance index and using a stochastic approximation algorithm. The algorithm possesses the capability of finding the stochastic gradient and adapting the control parameters in the direction of descent. A non-stationary Markov model is developed to investigate asymptotic convergence properties of the proposed genetic algorithm. It is proved that the proposed genetic algorithm would asymptotically converge. Numerical results based on the classical functions are obtained to show the potential of the proposed algorithm.