A Control Scheme Based on Online Delay Evaluation for a Class of Networked Control Systems

A control scheme that integrates control technology with communication technology to solve the delay problem is introduced for a class of networked control systems: Networked Half-Link Systems (NHLS). Concretely, we use the master-slave clock synchronization technology to evaluate the delays online, and then the LQ optimal control based on delays is adopted to stabilize the controlled plant. During the clock synchronization process, the error of evaluated delays is inevitably induced from the clock synchronization error, which will deteriorate the system performances, and even make system unstable in certain cases. Hence, the discussions about the clock error, and the related control analysis and design are also developed. Specifically, we present the sufficient conditions of controller parameters that guarantee the system stability, and a controller design method based on the error of delays is addressed thereafter. The experiments based on a CANbus platform are fulfilled, and the experimental results verify the previous analytic results finally.     

PeerMD： A P2P Molecular Dynamics Simulation Framework Based on Web Services

PeerMD, a P2P molecular dynamics simulation frame-work based on Web services is proposed. It utilizes rich free CPU time and network bandwidth of P2P networks to provide enough resources for dynamics simulation of bio-macromolecule, and has resolved the problem that it is difficult to interoperate between heterogeneous peers in P2P environment through Web services. Structure of PeerMD is given. Function, input and output of molecular dynamics simulation Web service are defined. Processes of publishing, discovering and invoking of molecular dynamics simulation Web service based on multi-level SuperPeer are given. A protocol system of PeerMD is implemented on a basic P2P platform JXTA, and experimental simulations of tumor necrosis fact alpha （TNF-α） and two mutations of it are executed on the protocol system. Simulation results show that PeerMD can speed up molecular dynamics simulation perfectly.     

Characterizing big data analytics workloads on POWER8 SMT processors

Big data analytics is emerging as one kind of the most important workloads in modern data cen-ters.Hence, it is of great interest to identify the method of achieving the best performance for big data analytics workloads running on state-of-the-art SMT ( simultaneous multithreading) processors, which needs comprehensive understanding to workload characteristics.This paper chooses the Spark workloads as the representative big data analytics workloads and performs comprehensive measure-ments on the POWER8 platform, which supports a wide range of multithreading.The research finds that the thread assignment policy and cache contention have significant impacts on application per-formance.In order to identify the potential optimization method from the experiment results, this study performs micro-architecture level characterizations by means of hardware performance counters and gives implications accordingly.     

Fault-Tolerant Design Techniques in A CMP Architecture

Single-chip multiprocessor ( CMP ) combined with the fault-tolerant (FT) techniques offers an ideal architecture to achieve high availability on the basis of sustaining high computing performance. FT design of a single-chip multiprocessor is described, including the techniques from hardware redundancy to software support and firmware strategy. The design aims at masking the influences of errors and automatically correcting the system states.     

基于CAN总线的分布式容错系统实现

针对一种基于控制器局域网(controller areanetwork,CAN)总线的多冗余计算机构成的实时控制网络,设计并实现了分布式容错计算系统。该系统采用软总线保证数据可靠传输,使用数据表决屏蔽节点计算与传感故障,通过同步保证数据一致,利用成员关系管理维护动态成员关系。该系统完全独立于控制应用程序,包括了分布式容错计算的典型功能,充分利用分布式系统的优势和CAN总线传输的特点,在保证系统实时性的基础上为应用程序提供容错计算服务。测试结果表明,在性能指标上能够满足某实时控制系统的要求。     

基于时滞状态反馈的D稳定鲁棒容错控制

针对时滞不确定受控对象,在状态反馈的基础上引入时滞状态反馈,基于Lyapunov稳定性理论和Riccati方程,给出线性离散一步时滞不确定系统对传感器失效具有D稳定鲁棒容错性需满足的一个充分条件和控制器的设计方法.利用该方法设计的闭环系统,在传感器发生故障时仍能保证稳定并具有一定的性能.仿真结果表明,引入时滞状态反馈后,系统的动态平稳性优于仅采用状态反馈的控制效果;经仿真结果比较,指出了时滞状态反馈增益矩阵的选取原则.进而对系统采用仅有一步时滞状态反馈的控制律,仿真结果证实了此方法的有效性.     

基于联邦滤波的多传感器主动容错估计方法

针对冗余传感器系统的容错问题,改进一种有反馈式联邦滤波的容错估计算法。通过残差法对硬故障进行检测,并隔离故障子系统的估计值。若全部子系统出现硬故障,取上一时刻的联邦滤波的最优值,通过系统状态方程计算得到系统状态量的预测值进行容错补偿。定义可信度因子,设计可信度单调性法则检测子系统软故障,并设计校正策略。大量仿真实验表明,改进后的反馈式联邦滤波的容错估计算法,可在保证系统估计精度的前提下,提高多传感器系统的容错性与鲁棒性。     

基于在线模糊聚类的故障检测和容错控制

提出了一种新的基于T-S模糊模型在线辨识的非线性系统的故障诊断与容错控制策略.在闭环控制中,根据在线产生的数据对T-S模糊模型进行辨识,当故障引起系统动态的结构性变化时,在线模糊聚类算法能够辨识出系统的重大改变并产生新的模糊规则描述系统新的动态,同时基于T-S模型的控制律也会做出相应的调整.分析了T-S模糊模型结构和参...     

具有容错机制的网格监控系统

提出了一种基于JMX实现远程监控的网格监控系统.重点论述具有容错机制的层次式系统架构的实现原理,系统采用模块化结构,通过备用管理节点的自动选择和实时在线启动实现容错机制.测试结果表明,可以减少单点故障造成的影响,提高系统的可靠性.     

基于T-S云模型的非线性网络化控制系统保性能鲁棒容错控制

基于T-S云模型对一类具有参数不确定的非线性网络化控制系统,考虑网络延时和丢包的影响,通过构造Lyapunov-Krasovskii泛函,推证出确保非线性网络化控制系统在执行器或传感器发生失效故障时具有保性能鲁棒容错性能的时滞依赖充分条件,并以求解线性或非线性矩阵不等式的方式给出控制器的设计方法.最后以一个仿真算例验证文中所述方法的可行性和有效性.     

Peer-to-Peer流媒体系统中容错机制的研究

基于分片策略和擦除纠错码技术的使用,提出和分析了一种面向P2P流媒体系统的新型容错机制,通过马尔可夫链模型来模拟系统的可靠性,可以得出系统冗余开销与可靠的系统规模之间的关系,结果表明：在Honeycomb系统有200个节点的情况下(即使是在单个节点很不可靠的情况下),只需要15.5％的冗余开销就可以获得系统较高的可靠性。     

基于LMI的网络化控制系统鲁棒容错控制器的设计

研究一类带有未知扰动的网络化控制系统的鲁棒容错控制问题.针对NCSs的时延和不同步对系统性能产生的影响,将时延的不确定性转换为系统状态方程系数矩阵的不确定性,将网络化控制系统的状态向量扩展为增广状态向量.基于Lyapunov稳定性理论和LMI方法,系统采用状态反馈控制策略,推证出保证闭环网络化控制系统在执行器或传感器发生失效故障时仍渐近稳定的充分条件,通过求解线性矩阵不等式组方便地得到容错控制器的设计方法.用仿真算例验证此种方法的可行性、有效性以及通用性.