全互连立方体网络在并行处理系统中的应用

提出一种应用于大规模并行处理系统的结点度等于常数的递归多级分层互连网络 ,称为全互连立方体网络 (fully connected cubic network,FCCN) .FCCN具有可扩展性好、延伸性能好等优点 .一个 m- FCCN可以由 8个(m - 1) - FCCN递归得到 ,FCCN网络的结点度与网络的规模大小无关等于常数 4,网络的直径和平均结点距离都与结点数的立方根成正比 .提出 FCCN中的简单路由算法 .并将 FCCN互连网络结构在大规模光电混合处理系统中进行应用 ,通过实际计算结果证明 FCCN具有比较高的并行处理效率     

Fully Distributed Nash Equilibrium Seeking for High-Order Players With Actuator Limitations

This paper explores the problem of distributed Nash equilibrium seeking in games, where players have limited knowledge on other players’ actions. In particular, the involved players are considered to be high-order integrators with their control inputs constrained within a pre-specified region. A linear transformation for players’ dynamics is firstly utilized to facilitate the design of bounded control inputs incorporating multiple saturation functions. By introducing consensus protocols with adaptive and time-varying gains, the unknown actions for players are distributively estimated. Then, a fully distributed Nash equilibrium seeking strategy is exploited, showcasing its remarkable properties: 1) ensuring the boundedness of control inputs; 2) avoiding any global information/parameters; and 3) allowing the graph to be directed. Based on Lyapunov stability analysis, it is theoretically proved that the proposed distributed control strategy can lead all the players’ actions to the Nash equilibrium. Finally, an illustrative example is given to validate effectiveness of the proposed method.      

Distributed control for a multi-evaporator air conditioning system

An autonomous hierarchical distributed control (AHDC) strategy is proposed for a building multi-evaporator air conditioning (ME A/C) system in this paper. The objectives are to minimize peak demand and energy costs, and to reduce communication resources, computational complexity and conservativeness while maintaining both thermal comfort and indoor air quality (IAQ) in acceptable ranges. The building consists of multiple connected rooms and zones. The proposed control strategy consists of two layers. The upper layer is an open loop optimizer, which only collects local measurement information and solves a distributed steady state resource allocation problem to autonomously and adaptively generate reference points, for low layer controllers. This is achieved by optimizing the demand and energy costs of a multi-zone building ME A/C system under a time-of-use (TOU) rate structure, while meeting the requirements of each zone’s thermal comfort and IAQ within comfortable ranges. The lower layer also uses local information to track the trajectory references, which are calculated by the upper layer, via a distributed model predictive control (DMPC) algorithm. The control strategy is distributed at both layers because they use only local information from the working zone and its neighbors. Simulation results are provided to illustrate the advantages of the designed control schemes.     

保留几何特征的散乱点云简化方法

针对散乱点云简化时经常丢失过多的几何特征,提出一种保留几何特征的简化方法。首先采用均匀栅格法划分点云空间;然后分别以点云中的数据点为球心构建包围球,并在包围球中查找数据点的K邻域;随后构造一个非负函数用于度量重建曲面在各点处的曲率,进而提取并保留点云中的特征点;最后根据法向量的内积阈值对包围球中的非特征点进行适度简化。实验结果表明该方法不仅能够充分保留点云中的几何特征,而且具有速度快的特点。     

Dynamic Sampling and Rendering of Algebraic Point Set Surfaces

Algebraic Point Set Surfaces (APSS) define a smooth surface from a set of points using local moving least‐squares (MLS) fitting of algebraic spheres. In this paper we first revisit the spherical fitting problem and provide a new, more generic solution that includes intuitive parameters for curvature control of the fitted spheres. As a second contribution we present a novel real‐time rendering system of such surfaces using a dynamic up‐sampling strategy combined with a conventional splatting algorithm for high quality rendering. Our approach also includes a new view dependent geometric error tailored to efficient and adaptive up‐sampling of the surface. One of the key features of our system is its high degree of flexibility that enables us to achieve high performance even for highly dynamic data or complex models by exploiting temporal coherence at the primitive level. We also address the issue of efficient spatial search data structures with respect to construction, access and GPU friendliness. Finally, we present an efficient parallel GPU implementation of the algorithms and search structures.     

基于InfiniBand的多链路mesh/torus大规模并行系统互连网络

在大规模并行系统中,系统级互连网络的设计至关重要.InfiniBand作为一种高性能交换式网络被广泛应用于大规模并行处理系统中.mesh/torus拓扑结构相较于目前普遍应用于InfiniBand网络的胖树拓扑结构拥有更好的性能与可扩展性.尽管如此,研究发现,用传统的mesh/torus拓扑结构构建InfiniBand互连网络存在诸多问题.分析了传统网络拓扑结构的缺陷,并提出了一种基于InfiniBand的多链路mesh/torus互连网络.这种改进型的拓扑结构通过充分利用交换机间的多链路可以获得比传统mesh/torus网络更高的带宽.另外,同时给出了与该网络拓扑结构相配套的高效路由算法.最后,通过网络仿真技术对提出的算法进行了评估,实验结果显示提出的路由算法相较于其他路由算法拥有更好的性能与可扩展性.     

Exploiting colored Petri nets to decide on permutation admissibility

In this work, we propose an innovative approach to investigate the admissibility of permutations to multistage interconnection networks—a challenging problem of switching theory. The proposed approach is centered upon modeling of multistage interconnection networks with colored Petri nets and use of Petri net analysis tools such as the unfolding technique and the invariants method. To assess the feasibility of the proposed approach we demonstrate that the complete unfoldings obtained in this work are polynomial in the problem size and employ an acyclic structure. The approach takes advantage of easy to use, yet extremely efficient, software tools.     

一类分布式系统的自适应跟踪和扰动抑制

研究直接自适应状态反馈控制策略解决一类有故障和摄动关联链接的分布式大系统渐进跟踪和扰动抑制问题. 根据特殊的分布式结构, 在所有关联故障因子, 关联通道摄动和子系统外部干扰的上界都未知下, 提出自适应率在线升级控制器参数. 基于自适应策略信息, 构造一类分布式状态反馈控制器自动补偿故障和摄动影响, 同时抑制外部扰动. 在关联通道有故障摄动和外部扰动情况下, 所提出的自适应鲁棒跟踪控制器可以保证所得自适应闭环大系统稳定, 及每个子系统渐进输出跟踪所对应参考信号. 最后由一个仿真例子评估所提技术的有效性.     

Improved genetic algorithm for VLSI floorplan design with non-slicing structure

Floorplan design is an important engineering problem. This problem can be modeled as a combinatorial optimization problem, in which a given set of rectangles in floorplan must be arranged. The goal is to find the arrangement with minimum area and minimum interconnection. Floorplans can be classified into slicing structure and non-slicing structure problems. In both problem types, it is difficult to obtain a true optimal solution in a time that could be used in practice. We proposed the new immune algorithm for optimization of the slicing structure problem in the past. In this paper, we focus on and analyze the improved points considered to be especially effective, and propose a new genetic algorithm for the non-slicing structure problem. The proposed method is compared with existing methods using well-known benchmark problems.     

Distributed adaptive robust tracking and model matching control with actuator faults and interconnection failures

In this paper, direct adaptive-state feedback control schemes are developed to solve the robust tracking and model matching control problem for a class of distributed large scale systems with actuator faults, faulty and perturbed interconnection links, and external disturbances. The adaptation laws are proposed to update the controller parameters on-line when all the eventual faults, the upper bounds of perturbations and disturbances are assumed to be unknown. Then a class of distributed state feedback controllers is constructed to automatically compensate the fault, perturbation and disturbance effects based on the information from adaptive schemes. The proposed distributed adaptive tracking controller can ensure that the resulting adaptive closed-loop large-scale system is stable and the tracking error decreases asymptotically to zero in the presence of uncertain faults of actuators and interconnections, perturbations in interconnection channels, and disturbances. The proposed adaptive design technique is finally evaluated in the light of a simulation example.     

传感器网络分布式事件触发多目标估计

本文主要研究无线传感器网络中目标数目已知且固定的一类分布式多目标跟踪问题,提出了一种完全分布式的基于事件触发的测量和通信策略使得每个节点在不需要全局信息的情况下实现估计误差和能量消耗之间的平衡.监测区域存在多个移动目标,传感器能否测量到单个目标由事件触发测量机制和节点的测量半径来综合决定.基于节点和邻居的信息采用k-means聚类算法来解决数据关联问题,同时提出了基于最小迹原则的一致性卡尔曼滤波算法.从理论上证明了该事件触发策略不仅在性能指标上优于基于时间触发的算法,而且在网络中如果存在节点对多目标协同可观,系统估计误差在均方意义下是稳定的.最后给出了仿真例子验证了该算法的有效性和可行性.     

Identification and decentralized adaptive control using dynamicalneural networks with application to robotic manipulators

Efficient implementation of a neural network-based strategy for the online adaptive control of complex dynamical systems characterized by an interconnection of several subsystems (possibly nonlinear) centers on the rapidity of the convergence of the training scheme used for learning the system dynamics. For illustration, in order to achieve a satisfactory control of a multijointed robotic manipulator during the execution of high speed trajectory tracking tasks, the highly nonlinear and coupled dynamics together with the variations in the parameters necessitate a fast updating of the control actions. For facilitating this requirement, a multilayer neural network structure that includes dynamical nodes in the hidden layer is proposed, and a supervised learning scheme that employs a simple distributed updating rule is used for the online identification and decentralized adaptive control. Important characteristic features of the resulting control scheme are discussed and a quantitative evaluation of its performance in the above illustrative example is given.     

Consensus-based distributed cooperative learning control for a group of discrete-time nonlinear multi-agent systems using neural networks

This paper focuses on the cooperative learning capability of radial basis function neural networks in adaptive neural controllers for a group of uncertain discrete-time nonlinear systems where system structures are identical but reference signals are different. By constructing an interconnection topology among learning laws of NN weights in order to share their learned knowledge on-line, a novel adaptive NN control scheme, called distributed cooperative learning control scheme, is proposed. It is guaranteed that if the interconnection topology is undirected and connected, all closed-loop signals are uniform ultimate bounded and tracking errors of all systems can converge to a small neighborhood around the origin. Moreover, it is proved that all estimated NN weights converge to a small neighborhood of their common optimal value along the union of all state trajectories, which means that the estimated NN weights reach consensus with a small consensus error. Thus, all learned NN models have the better generalization capability than ones obtained by the deterministic learning method. The learned knowledge is also adopted to control a class of uncertain systems with the same structure but different reference signals. Finally, a simulation example is provided to verify the effectiveness and advantages of the distributed cooperative learning control scheme developed in this paper.     

基于能量整形方案实现具有通讯时滞欠驱动Euler-Lagrange网络的一致性

本文主要基于能量整形方案研究具有通讯时滞网络化欠驱动Euler-Lagrange (EL)系统的一致性问题,通过利用阻尼注入和互连分配的无源控制(PBC)技术,在有向连通网络拓扑下提出了一个简单的分布式协议,来实现在无引导者和有引导者-跟随者两种情形下欠驱动EL网络的一致性. 本文提出的一致性能量整形方案的主要特点是有机地整合了系统欠驱动和驱动部分以及控制器三部分能量作为整个系统的总能量,这个总能量被利用作为一个合适的Lyapunov函数,它能够充分确保网络化欠驱动EL系统达到所期望的分布式一致性. 最后,通过由欠驱动EL网络所描述柔性关节机械臂系统的数值模拟,来分析通讯时滞对一致性的效应和验证所提出控制算法的正确性.     

A comprehensive analytical model of interconnection networks in large‐scale cluster systems

The trends in parallel processing system design and deployment have been toward networked distributed systems such as cluster computing systems. Since the overall performance of such distributed systems often depends on the efficiency of their communication networks, performance analysis of the interconnection networks for such distributed systems is paramount. In this paper, we develop an analytical model, under non‐uniform traffic and in the presence of communication locality, for the m‐port n‐tree family interconnection networks commonly employed in large‐scale cluster computing systems. We use the proposed model to study two widely used interconnection networks flow control mechanism namely the wormhole and store&forward. The proposed analytical model is validated through comprehensive simulation. The results of the simulation demonstrated that the proposed model exhibits a good degree of accuracy for various system organizations and under different working conditions. Copyright © 2007 John Wiley & Sons, Ltd.     

高分辨率SAR影像同名点自动匹配技术

由于SAR工作原理跟光学遥感成像技术截然不同，而经典同名点匹配方法已不能完全满足SAR自动配准需要。为了快速有效地进行SAR影像配准，基于SAR影像自动配准技术的发展，提出了一种新颖的高分辨率SAR影像同名点自动匹配算法，该算法首先创建金字塔影像，同时在金字塔影像上回溯搜索，以确定初始变换函数类型及相应的变换参数；然后通过分层回溯逐层加密控制点来解求最佳变换函数类型及相应变换参数；最后在原始影像分辨率下修正同名点坐标，以获取最终匹配同名点对。这种分层回溯策略不仅很好地解决了同名点搜索计算复杂度问题，使获取的同名点对分布更趋均匀、精度更高，而且能确保每层变换函数达到全局最小二乘最优。另外，以高分辨率SAR影像为实验数据进行的实验结果表明，该同名点搜索算法不仅计算时间可由221s缩短至34s，而且可达到0．284636 pixels的配准精度。     

Quantized tracking control for a multi‐agent system with high‐order leader dynamics

This paper presents a study of a tracking control problem for a multi‐agent system with an active leader and quantized communication constraints. We first design a discrete‐time distributed estimator‐based tracking control for each follower‐agent and analyze the tracking convergence with the help of the Riccati equation and common Lyapunov function when the communication channel is perfect and the interconnection topology is time‐varying. Then a stochastic quantization strategy is applied to model the information communication in the agent coordination and the corresponding solution to the tracking control problem is also given. Finally, a numerical example is given to illustrate the tracking control algorithm. Copyright © 2010 John Wiley and Sons Asia Pte Ltd and Chinese Automatic Control Society     

Adaptive multi-agent containment control with multiple parametric uncertain leaders

In this paper, an adaptive containment control is considered for a class of multi-agent systems with multiple leaders containing parametric uncertainties. The agents are heterogeneous though their dynamics have the same relative degree and are minimum phase, while the interconnection topology is described by a general directed graph. A distributed containment control is proposed for the agents to enter the moving convex set spanned by the leaders, based on an adaptive internal model and a recursive stabilization control law.     

Q-PSO: Fast Quaternion-Based Pose Estimation from RGB-D Images

This paper investigates the leader–follower formation control problem for a group of networked nonholonomic mobile robots that are subject to bounded time-varying communication delays and an asynchronous clock. First we convert the formation control problem into a trajectory tracking problem, and then a fully distributed unified control framework based on the receding horizon control is implemented to converge the tracking errors. By adding an auxiliary acceleration term into the receding horizon controller, the framework is able to solve the impractical velocity jump problem. Considering the time-varying delays, the timing and order features of the messages are utilized to guarantee their logical correctness. To compensate for the delay effect, an improved control framework that exploits the predictability of the receding horizon controller is proposed. The asynchronous clock problem, which makes the communication delay unmeasurable, is studied. We give a definition of the syn point that is inspired from investigation of the property that messages are received out of order in a bounded time-varying delayed network. A novel method that detects the occurrence of syn points is integrated into the control framework to solve the asynchronous clock problem. Finally the effectiveness of the proposed approaches is demonstrated in the Player/Stage simulation environment.     

State analytical predictor for systems with multiple time delays

An approach for the design of a dead-time compensator for processes with time delays is presented. The proposed algorithm deals with multivariable state-space models instead of input-output models and utilizes not only a single delay but also distributed delays in the feedback loop. Both the continuous-time and discrete-time versions of the algorithm are derived. The design strategy of the controller is based on the approach of LQG design. It is incorporated into the dead-time compensation technique of the analytical predictor, which uses the process model directly to predict the effect of input variables on the process outputs. By introducing an integral action into the observation system, the steady-state observation error of the inaccessible load inputs converges to zero. This permits us to perform disturbance prediction and compensation within a single design. Theoretical analysis shows that this control strategy fully removes time delay elements from a system characteristic equation in the ideal situation of a system without model-plant mismatch and/or noise measurements. By the estimation and prediction of the unknown plant-input disturbances, the capability for rejecting input disturbances has also been improved. The control structure developed is shown to have the same closed-loop relationships as the linear quadratic feedback structure for a system without lime delays. The potential benefits of the proposed controller are demonstrated by the control problem of an industrial grinding system studied previously by Niemi et al. (1982) and Ylinen et al. (1987), which is a challenging problem for MPC-type algorithms.