一类高阶线性系统的分层控制

近年来, 一种基于系统层次结构的分层控制方法发展了起来, 它主要针对的是高阶复杂系统. 为了分析和控制这类系统, 我们首先根据需要构建一个合适的低阶抽象系统, 然后在此基础上分析和设计原始系统控制律, 即所谓的接口函数. 本文将此方法用于一类高阶线性系统的稳定控制问题, 并得到了相应的控制律. 最后, 算例部分的仿真结果说明了这种方法的有效性.     

基于本体的分层抽象模型

KRA表示模型给出了对物理世界W进行形式化建模的一般框架。在扩展后的广义KRA模型中引入本体概念,将抽象对象库扩展为本体类,在本体类中实现KRA模型框架中的感知层、语言层和理论层的知识共享和重用,简化了KRA模型的表示。分别定义了作用在物理世界W和本体类上的3种本体抽象算子(集):基本本体抽象算子、实体本体抽象算子集和连接本体抽象算子集,并给出了本体类之间的映射关系,通过这种映射关系能够实现模型抽象和模型还原。对本体类的抽象度进行形式化的定义,并给出了相关定理,指出在这样的框架中利用抽象映射可以自动构建物理世界W的抽象度不同的模型。     

多机并发系统中通信模型分层抽象的方法

针对多机并发系统的复杂性,为了辅助用户能从多个角度和层次全面地理解并发系统,就需要逆向产生出能够反映软件系统框架结构的高层架构。基于此本文以进程为边界,提出了一种分层抽取多机并发系统通信模型的方法。此方法基于反射和开放编译的植入机制来获取所需要的动态信息,在此基础上运用分层抽象的策略,分别从系统、节点、进程三个层次对多机并发系统的通信结构和设计结构进行逆向恢复,最后对该方法进行系统的实验研究。结果表明,分层抽象所得到并发系统的通信模型能够正确、有效地反映系统设计时的高层架构关系。     

机载无线电导航系统级接口仿真平台研究

根据机载无线电导航相关组件的特点,设计了基于ARINC规范的系统级接口仿真平台,能够完成对相关机载无线电导航组件的ICD(接口控制文档)进行接口数据仿真;首先对该仿真平台的功能和特点进行了介绍,然后给出仿真平台具体的硬件和软件实现方法;创新之处在于仿真平台设计过程中将待仿真的ICD与具体仿真软件分离,除此之外,仿真平台部分数据交互采用面向无连接的通信协议UDP,在一定上增加了仿真平台的通用性和布局灵活性;目前该仿真平台已成功应用于我国某型飞机的导航系统仿真实验中。     

Hierarchical control system design using approximate simulation

In this paper, we present a new approach for hierarchical control based on the recent notions of approximate simulation and simulation functions, a quantitative version of the simulation relations. Given a complex system that needs to be controlled and a simpler abstraction, we show how the knowledge of a simulation function allows us to synthesize hierarchical control laws by first controlling the abstraction and then lifting the abstract control law to the complex system using an interface. For the class of linear control systems, we give an effective characterization of the simulation functions and of the associated interfaces. This characterization allows us to use algorithmic procedures for their computation. We show how to choose an abstraction for a linear control system such that our hierarchical control approach can be used. Finally, we show the effectiveness of our approach on an example.     

Knowledge-based spatiotemporal linear abstraction

We present a theoretical framework and a case study for reusing the same conceptual and computational methodology for both temporal abstraction and linear (unidimensional) space abstraction, in a domain (evaluation of traffic-control actions) significantly different from the one (clinical medicine) in which the method was originally used. The method, known asknowledge-based temporal abstraction, abstracts high-level concepts and patterns from time-stamped raw data using a formal theory of domain-specific temporal-abstraction knowledge. We applied this method, originally used to interpret time-oriented clinical data, to the domain of traffic control, in which the monitoring task requires linear pattern matching along both space and time. First we reused the method for creation of unidimensional spatial abstractions over highways, given sensor measurements along each highway measured at the same time point. Second, we reused the method to create temporal abstractions of the traffic behaviour, for the same space segments, but during consecutive time points. We defined the corresponding temporal-abstraction and spatial-abstraction domain-specific knowledge. Our results suggest that (1) the knowledge-base temporal-abstraction method is reusable over time and unidimensional space as well as over significantly different domains; (2) the method can be generalised into a knowledge-based linear-abstraction method, which solves tasks requiring abstraction of data along any linear distance measure; and (3) a spatiotemporal-abstraction method can be assembled, from two copies of the generalised method and a spatial-decomposition mechanism, and is applicable to tasks requiring abstraction of time-oriented data into meaningful spatiotemporal patterns over a linear, decomposable space, such as traffic over a set of highways.     

基于分层存储理论模型的近似字符串匹配并行算法研究

CUDA (Compute Unified Device Architecture)是一种重要的并行处理架构,但其具有相对复杂的线程管理机制和多重存储模块,从而使得基于CUDA的算法时间复杂度很难量化.针对这一问题,提出了一种分层存储理论模型—HMM (Hierarchical Memory Machine)模型,该模型所具有的分层存储结构可以有效地描述图形处理单元设备不同存储模块的物理特性,因此非常适用于对CUDA算法时间复杂度的量化评估.作为HMM模型的应用实例,文章提出了一种基于HMM模型的并行近似字符串匹配算法,并给出了相应算法时间复杂度的计算过程.与串行算法相比,该算法可以获得60倍以上的加速比.     

基于系统中心本体的分层抽象模型

本体类分层关系的确定,使得能够利用不同抽象度本体类之间的抽象映射自动构建物理世界的分层模型。根据本体类型的不同将已有的本体类定义为对象本体类(Object based Ontology Class),并对流片段(Flow Fragment)的概念加以扩展,提出以系统为中心的流本体类(Flow-based Ontology Class)的概念。定义了流片段的行为不可区分性,从两个不同的方向讨论了流本体类的分层表示:一是对构成流本体类的流片段的不同抽象度分层扩展;二是对构成流本体类的不可区分的流片段的合并分层扩展。给出了流本体类的分层过程,定义了系统级的映射关系,即流片段之间的抽象映射,指出该映射为基于系统中心本体的模型抽象分层过程提供了直接的转换运算。并且,流本体类的分层关系的确定也为基于系统中心本体的模型设计任务提供了可共享和重用的机制。     

基于拓扑抽象的高性能网络模拟方法

为提升网络模拟性能,提出了基于拓扑抽象的网络模拟方法。该方法对大规模网络拓扑进行抽象,形成小规模网络拓扑;通过对该小规模拓扑的模拟结果进行分析,形成原始大规模拓扑的模拟结果。理论证明了该方法能保证路由器队列长度、丢包率、数据包转发时延、TCP流量速率等瞬时网络行为参数的模拟真实性。通过基于NS2的实验结果表明,当拓扑抽象度达到16时,相对于传统的网络模拟,该方法降低模拟运行时间98%以上,而网络行为模拟结果仍具有较高的真实性。     

Abstractions for hybrid systems

We present a procedure for constructing sound finite-state discrete abstractions of hybrid systems. This procedure uses ideas from predicate abstraction to abstract the discrete dynamics and qualitative reasoning to abstract the continuous dynamics of the hybrid system. It relies on the ability to decide satisfiability of quantifier-free formulas in some theory rich enough to encode the hybrid system. We characterize the sets of predicates that can be used to create high quality abstractions and we present new approaches to discover such useful sets of predicates. Under certain assumptions, the abstraction procedure can be applied compositionally to abstract a hybrid system described as a composition of two hybrid automata. We show that the constructed abstractions are always sound, but are relatively complete only under certain assumptions.     

Programming by successive refinement of data abstractions

The Model Programming Language implements a form of data abstraction that has been used in a large programming project, the Demos Operating System. The use of the abstraction mechanism suggests a particular programming style that has evolved over an extensive period of gaining experience with the language. The programming style and Model's approach to data abstraction are both documented here using an example designed to illustrate several of the more important issues. The goal of the paper is to demonstrate a programming style and an approach to data abstraction in a programming language that has proved useful in a significant systems programming application.     

石油化工过程动态仿真的递阶方法

石油化工过程动态仿具的递阶方法融合了面向方程和面向模块方法各自的优点，将流量变化等工艺过程中主要的“快过程”用面向方程的方法求角，将组分变化，相平衡、化学反应等“慢过程”按设备组成相对独立的模块，用面向模块的方法求解。工艺过程与控制系统分别提高了仿真软件的可移植性，变步长仿真方法提高了仿真效率。     

利用输出信息的变结构控制设计方法*

本文对有指定极点集的切换函数给出了存在条件及其一般形式,提出了利用最佳输出逼近和极点配置设计没态的方法,改进和推广了文［１］的结果。     

大型DCS轮机仿真I/O接口系统研究

I／O接口系统的性能直接关系到轮机仿真系统的实时性和稳定性，需要对其进行深入研究。文章介绍了作者自主研制开发的分布式轮机仿真I／O接口系统的组成。讨论了其各硬件板卡的设计，分析了该系统的数据通信原理。系统实际应用表明，该系统特别适用于大型、复杂的系统仿真。     

基于目标检测及跟踪的视频摘要技术研究

为了能够在不丢失有用信息的前提下对海量监控视频进行摘要,提出一种基于目标检测及跟踪的视频摘要技术。首先根据混合高斯模型对视频进行背景建模,获得监控视频的背景;然后通过背景减法对运动目标进行检测,得到前景运动目标;其次对检测到的运动目标采用分层关联的思想进行目标跟踪,得到运动目标的完整信息;最后将这些运动目标及视频背景重新组装成摘要视频。实验结果表明,该方法能够有效地对监控视频进行浓缩,形成的摘要视频能够完整地保存原有视频信息,减少了存储空间,降低了成本,方便了相关人员及时获取有用信息,提高了工作效率。     

Hierarchical computation of decentralized gains for interconnected systems

In this paper, the problem of computing the decentralized gains for linear, interconnected dynamical systems is treated as a constrained optimization problem. A two-level hierarchical computational structure is developed to determine the decentralized gains. At the lower level, two Lyapunov equations are solved whilst a predictive-corrective routine is used at the higher level. Simulation results of two control examples show that our two-level structure needs significantly less computational effort than previous approaches.