基于Radau伪谱法的非线性最优控制问题的收敛性

在过去的10年里,伪谱方法(如Legendre伪谱法、Gauss伪谱法、Radau伪谱法)逐步成为求解不同领域中非线性最优控制问题的一种高效、灵活的数值解法.本文从最优控制问题解的存在性、收敛性以及解的可行性3个方面对采用Radau伪谱法求解一般非线性最优控制问题解的收敛性进行研究.证明了原最优控制问题的离散解存在、存在收敛到原最优控制问题解上的离散解和离散形式的收敛解是原最优控制问题的最优解.在此基础上,证明了Radau伪谱法的收敛性.本文结论与现有文献相比,去掉了一些必要条件,更适合一般的非线性时不变系统.     

A quasi-newton differential dynamic programming algorithm for discrete-time optimal control

We develop a quasi-Newton differential dynamic programming algorithm (QDDP) for discrete-time optimal control problems. In the spirit of dynamic programming, the quasi-Newton approximations are performed in a stagewise manner. We establish the global convergence of the method and also show a superlinear convergence rate. Among other advantages of the QDDP method, second derivatives need not be calculated. In theory, the computational effort of each recursion grows proportionally to the number of stages N, whereas with conventional quasi-Newton techniques which do not take advantage of the optimal control problem structure, the growth is as N². Computational results are also reported.     

Minimum Bayes error features for visual recognition

The design of optimal feature sets for visual classification problems is still one of the most challenging topics in the area of computer vision. In this work, we propose a new algorithm that computes optimal features, in the minimum Bayes error sense, for visual recognition tasks. The algorithm now proposed combines the fast convergence rate of feature selection (FS) procedures with the ability of feature extraction (FE) methods to uncover optimal features that are not part of the original basis function set. This leads to solutions that are better than those achievable by either FE or FS alone, in a small number of iterations, making the algorithm scalable in the number of classes of the recognition problem. This property is currently only available for feature extraction methods that are either sub-optimal or optimal under restrictive assumptions that do not hold for generic imagery. Experimental results show significant improvements over these methods, either through much greater robustness to local minima or by achieving significantly faster convergence.     

A pseudospectral method for the optimal control of constrained feedback linearizable systems

We consider the optimal control of feedback linearizable dynamical systems subject to mixed state and control constraints. In general, a linearizing feedback control does not minimize the cost function. Such problems arise frequently in astronautical applications where stringent performance requirements demand optimality over feedback linearizing controls. In this paper, we consider a pseudospectral (PS) method to compute optimal controls. We prove that a sequence of solutions to the PS-discretized constrained problem converges to the optimal solution of the continuous-time optimal control problem under mild and numerically verifiable conditions. The spectral coefficients of the state trajectories provide a practical method to verify the convergence of the computed solution. The proposed ideas are illustrated by several numerical examples.     

Improvement of harmony search algorithm by using statistical analysis

In this article, we suggest a new method to improve the harmony search meta-heuristic algorithm. Several approaches are presented for improving the harmony search algorithm. These approaches consider different values for initial parameters in each optimization problem. Differences between the proposed algorithm and the harmony search algorithm are as follows. First, we add a new step to create a new harmony vector, which increases the accuracy and convergence rate and reduces the impact of the initial parameters in achieving an optimal solution. Second, we set introduce a parameter called bandwidth (bw), which is an important factor with great influence on the convergence rate toward optimal solutions. To prove the efficiency and robustness of the proposed algorithm, we argument about statistical analysis of proposed algorithm and examine it through a variety of optimization problems, including constrained and unconstrained functions, mathematical problems with high dimensions and engineering and reliability problems. In all of these problems, the convergence rate and accuracy of the answer are equal to or better than other methods. In addition, in our proposed method, the effect of initial parameters has been reduced with respect to the optimal solution.     

Modified alternating direction-implicit iteration method for linear systems from the incompressible Navier–Stokes equations

In order to solve the large sparse systems of linear equations arising from numerical solutions of two-dimensional steady incompressible viscous flow problems in primitive variable formulation, Ran and Yuan [On modified block SSOR iteration methods for linear systems from steady incompressible viscous flow problems, Appl. Math. Comput. 217 (2010), pp. 3050–3068] presented the block symmetric successive over-relaxation (BSSOR) and the modified BSSOR iteration methods based on the special structures of the coefficient matrices. In this study, we present the modified alternating direction-implicit (MADI) iteration method for solving the linear systems. Under suitable conditions, we establish convergence theorems for the MADI iteration method. In addition, the optimal parameter involved in the MADI iteration method is estimated in detail. Numerical experiments show that the MADI iteration method is a feasible and effective iterative solver.     

A domain-specific knowledge-based heuristic for the Blocks Relocation Problem

The Blocks Relocation Problem consists in minimizing the number of movements performed by a gantry crane in order to retrieve a subset of containers placed into a bay of a container yard according to a predefined order. A study on the mathematical formulations proposed in the related literature reveals that they are not suitable for its solution due to their high computational burden. Moreover, in this paper we show that, in some cases, they do not guarantee the optimality of the obtained solutions. In this regard, several optimization methods based on the well-known A¹ search framework are introduced to tackle the problem from an exact point of view. Using our A¹ algorithm we have corrected the optimal objective function value of 17 solutions out of 45 instances considered by Caserta et al. (2012) [4]. In addition, this work presents a domain-specific knowledge-based heuristic algorithm to find high-quality solutions by means of short computational times. It is based on finding the most promising positions into the bay where to relocate those containers that are currently located on the next one to be retrieved, in such a way that, they do not require any additional relocation operation in the future. The computational tests indicate the higher effectiveness and efficiency of the suggested heuristic when solving real-world scenarios in comparison with the most competitive approaches from the literature.     

Density estimation and random variate generation using multilayernetworks

In this paper we consider two important topics: density estimation and random variate generation. We present a framework that is easily implemented using the familiar multilayer neural network. First, we develop two new methods for density estimation, a stochastic method and a related deterministic method. Both methods are based on approximating the distribution function, the density being obtained by differentiation. In the second part of the paper, we develop new random number generation methods. Our methods do not suffer from some of the restrictions of existing methods in that they can be used to generate numbers from any density provided that certain smoothness conditions are satisfied. One of our methods is based on an observed inverse relationship between the density estimation process and random number generation. We present two variants of this method, a stochastic, and a deterministic version. We propose a second method that is based on a novel control formulation of the problem, where a "controller network" is trained to shape a given density into the desired density. We justify the use of all the methods that we propose by providing theoretical convergence results. In particular, we prove that the L(infinity) convergence to the true density for both the density estimation and random variate generation techniques occurs at a rate O((log log N/N)((1-epsilon)/2)) where N is the number of data points and epsilon can be made arbitrarily small for sufficiently smooth target densities. This bound is very close to the optimally achievable convergence rate under similar smoothness conditions. Also, for comparison, the (2) root mean square (rms) convergence rate of a positive kernel density estimator is O(N(-2/5)) when the optimal kernel width is used. We present numerical simulations to illustrate the performance of the proposed density estimation and random variate generation methods. In addition, we present an extended introduction and bibliography that serves as an overview and reference for the practitioner.     

Algorithms for coplanar camera calibration

Abstract. Coplanar camera calibration is the process of determining the extrinsic and intrinsic camera parameters from a given set of image and world points, when the world points lie on a two-dimensional plane. Noncoplanar calibration, on the other hand, involves world points that do not lie on a plane. While optimal solutions for both the camera-calibration procedures can be obtained by solving a set of constrained nonlinear optimization problems, there are significant structural differences between the two formulations. We investigate the computational and algorithmic implications of such underlying differences, and provide a set of efficient algorithms that are specifically tailored for the coplanar case. More specifically, we offer the following: (1) four algorithms for coplanar calibration that use linear or iterative linear methods to solve the underlying nonlinear optimization problem, and produce sub-optimal solutions. These algorithms are motivated by their computational efficiency and are useful for real-time low-cost systems. (2) Two optimal solutions for coplanar calibration, including one novel nonlinear algorithm. A constraint for the optimal estimation of extrinsic parameters is also given. (3) A Lyapunov type convergence analysis for the new nonlinear algorithm. We test the validity and performance of the calibration procedures with both synthetic and real images. The results consistently show significant improvements over less complete camera models. Received: 30 September 1998 / Accepted: 12 January 2000     

Consistent approximation of a nonlinear optimal control problem with uncertain parameters

This paper focuses on a non-standard constrained nonlinear optimal control problem in which the objective functional involves an integration over a space of stochastic parameters as well as an integration over the time domain. The research is inspired by the problem of optimizing the trajectories of multiple searchers attempting to detect non-evading moving targets. In this paper, we propose a framework based on the approximation of the integral in the parameter space for the considered uncertain optimal control problem. The framework is proved to produce a zeroth-order consistent approximation in the sense that accumulation points of a sequence of optimal solutions to the approximate problem are optimal solutions of the original problem. In addition, we demonstrate the convergence of the corresponding adjoint variables. The accumulation points of a sequence of optimal state-adjoint pairs for the approximate problem satisfy a necessary condition of Pontryagin Minimum Principle type, which facilitates assessment of the optimality of numerical solutions.     

Adaptive control for manufacturing systems using infinitesimalperturbation analysis

Combining infinitesimal perturbation analysis (IPA) with stochastic approximation gives identification algorithms to estimate the optimal threshold value for failure-prone manufacturing systems consisting of one machine producing one part type. Two adaptive control schemes are proposed. The adaptive control schemes do not require the knowledge of the distribution functions of the up and down times. Under some appropriate conditions, the strong consistency, as well as the convergence rates, of the identification algorithms and the cost function is established for the adaptive control schemes. In particular, it is shown that central limit theorems hold for the identification algorithms     

Investigation of Commutative Properties of Discontinuous Galerkin Methods in PDE Constrained Optimal Control Problems

The aim of this paper is to investigate commutative properties of a large family of discontinuous Galerkin (DG) methods applied to optimal control problems governed by the advection-diffusion equations. To compute numerical solutions of PDE constrained optimal control problems there are two main approaches: optimize-then-discretize and discretize-then-optimize. These two approaches do not always coincide and may lead to substantially different numerical solutions. The methods for which these two approaches do coincide we call commutative. In the theory of single equations, there is a related notion of adjoint or dual consistency. In this paper we classify DG methods both in primary and mixed forms and derive necessary conditions that can be used to develop new commutative methods. We will also derive error estimates in the energy and L ² norms. Numerical examples reveal that in the context of PDE constrained optimal control problems a special care needs to be taken to compute the solutions. For example, choosing non-commutative methods and discretize-then-optimize approach may result in a badly behaved numerical solution.     

Optimal remapping in dynamic bulk synchronous computations via a stochastic control approach

A bulk synchronous computation proceeds in phases that are separated by barrier synchronization. For dynamic bulk synchronous computations that exhibit varying phase-wise computational requirements, remapping at runtime is an effective approach to ensure parallel efficiency. The paper introduces a novel remapping strategy for computations whose workload changes can be modeled as a Markov chain. The use of a Markovian model allows us to treat statistical dependence and more complex structure than the usual independent identically distributed random variable assumptions. Our models are quite general and we do not need to impose conditions on the dynamics of the underlying process other than the transition probability matrix. It is shown that optimal remapping can be formulated as a binary decision process: remap or not at a given synchronizing instant. The optimal strategy is then developed for long lasting computations by employing optimal stopping rules in a stochastic control framework. The existence of optimal controls is established. Necessary and sufficient conditions for the optimality are obtained. Furthermore, a policy iteration algorithm is devised to reduce computational complexity and enhance fast convergence to the desired optimal control.     

Reliability-based design optimization with cooperation between support vector machine and particle swarm optimization

Reliability-based design optimization (RBDO) is concerned with designing an engineering system to minimize a cost function subject to the reliability requirement that failure probability should not exceed a threshold. Conventional RBDO methods are less than satisfactory in dealing with discrete design parameters and complex limit state functions (nonlinear and non-differentiable). Methods that are flexible enough to address the concerns above, however, come at a high computational cost. To enhance computational efficiency without sacrificing model flexibility, we propose a new RBDO framework: PS2, which combines Particle Swarm Optimization (PSO), Support Vector Machine (SVM), and Subset Simulation (SS). SS can efficiently estimate small failure probabilities, based on which SVM is adopted to evaluate the reliability of candidate solutions using binary classification. PSO is employed to solve the discrete optimization problem. Primary emphasis is placed upon the cooperation between SVM and PSO. The cooperation is mutually beneficial since the SVM classifier helps PSO evaluate the feasibility of solutions with high efficiency while the optimal solutions obtained by PSO assist in retraining the SVM classifier to attain better accuracy. The PS2 framework is implemented to find the optimal design of a ten-bar truss, whose component sizes are selected from a commercial standard. The reliability constraints are non-differentiable with two failure modes: yield stress and buckling stress. The interactive process between PSO and SVM contributes greatly to the success of the PS2 framework. It is shown that in various trials the PS2 framework consistently outperforms both the double-loop and single-loop approaches in terms of computational efficiency, solution quality, and model flexibility.     

Finite sample properties of system identification methods

In this paper we study the quality of system identification models obtained using the standard quadratic prediction error criterion for a general linear model class. The main feature of our results is that they hold true for a finite data sample and they are not asymptotic. The main theorems bound the difference between the expected value of the identification criterion evaluated at the estimated parameters and at the optimal parameters. The bound depends naturally on the model and system order, the pole locations, and the noise variance, and it shows that although these variables often do not enter in asymptotic convergence results, they do play an important role when the data sample is finite.     

An evolutionary algorithm approach to the share of choices problem in the product line design

In this paper an evolutionary algorithm (EA) approach for solving the share of choices problem in the design of a line of substitute products is presented. Because this problem is NP-hard, we are forced to use heuristic methods, which do not guarantee obtaining the optimal solution. In order to see whether there exists any computational advantage of the EA approach, it is compared to the beam search (BS) heuristic method. An extensive comparative computational study is performed. The solutions found by the EA are close to optimal. Moreover, in most cases the EA obtains a better solution than that found by the BS method. In most problem sizes the BS method is faster than the EA. However, the CPU time needed by the EA is very reasonable. Moreover, the EA can be used as an effective second step to the BS method.     

ANASA-a stochastic reinforcement algorithm for real-valued neuralcomputation

This paper introduces ANASA (adaptive neural algorithm of stochastic activation), a new, efficient, reinforcement learning algorithm for training neural units and networks with continuous output. The proposed method employs concepts, found in self-organizing neural networks theory and in reinforcement estimator learning algorithms, to extract and exploit information relative to previous input pattern presentations. In addition, it uses an adaptive learning rate function and a self-adjusting stochastic activation to accelerate the learning process. A form of optimal performance of the ANASA algorithm is proved (under a set of assumptions) via strong convergence theorems and concepts. Experimentally, the new algorithm yields results, which are superior compared to existing associative reinforcement learning methods in terms of accuracy and convergence rates. The rapid convergence rate of ANASA is demonstrated in a simple learning task, when it is used as a single neural unit, and in mathematical function modeling problems, when it is used to train various multilayered neural networks.     

Comparisons of some improving strategies on MOPSO for multi-objective (r, Q) inventory system

This paper presents comparisons of some recent improving strategies on multi-objective particle swarm optimization (MOPSO) algorithm which is based on Pareto dominance for handling multiple objective in continuous review stochastic inventory control system. The complexity of considering conflict objectives such as cost minimization and service level maximization in the real-world inventory control problem needs to employ more exact optimizers generating more diverse and better non-dominated solutions of a reorder point and order size system. At first, we apply the original MOPSO employed for the multi-objective inventory control problem. Then we incorporate the mutation operator to maintain diversity in the swarm and explore all the search space into the MOPSO. Next we change the leader selection strategy used that called geographically-based system (Grids) and instead of that, crowding distance factor is also applied to select the global optimal particle as a leader. Also we use ε-dominance concept to bound archive size and maintain more diversity and convergence in the MOPSO for optimizing the inventory control problem. Finally, the MOPSO algorithms created using these strategies are evaluated and compared with each other in terms of some performance metrics taken from the literature. The results indicate that these strategies have significant influences on computational time, convergence, and diversity of generated Pareto optimal solutions.     

Approximate discrete maximum principle for the difference approximation of optimal periodic control problems

Distinctive features of the difference approximation for optimal periodic control (OPC) problems are considered. It is shown that the convergence conditions on approximate solutions imply consistency conditions on the time grid step and the accuracy of the periodicity constraint approximation such that the approximate discrete maximum principle holds for a sequence of discretized problems. A specific form of this principle when applied to the difference approximation method for OPC problems is analysed and computational methods are discussed. Application of this approach to a class of systems non-linearly dependent on control variables, and occurring in the chemical industry, is suggested.     

Two extragradient methods for generalized mixed equilibrium problems, nonexpansive mappings and monotone mappings

In this paper, we introduce iterative schemes based on the extragradient method for finding a common element of the set of solutions of a generalized mixed equilibrium problem and the set of fixed points of an infinite (a finite) family of nonexpansive mappings and the set of solutions of a variational inequality problem for a monotone, Lipschitz continuous mapping. We obtain some weak convergence theorems for the sequences generated by these processes in Hilbert spaces. The results in this paper generalize, extend and unify some well-known weak convergence theorems in the literature.