基于神经网络非线性系统辨识和控制的研究

本文提出了由静态的前馈网络和稳定的滤波器构成的非线性系统的辨识模型，在神经网络固有的逼近误差存在的情况下，从理论上讨论了神经网络应用于辨识控制过程中系统的稳定性问题，最后研究了在非线性系统的轨迹跟踪过程中增加滑动控制来偿神经网络的逼近误差，从而提高系统跟踪性能。     

K-Dimensional Optimal Parallel Algorithm for the Solution of a General Class of Recurrence Equations

1IntroductionAlgorithmshavebeenproposedtosolvelinearrecurrencesinparallell1-13].Someofthemsupposeunlimitednumberofprocessorsbeingusedwhileothersuselimitednumberofprocessors.P-M.KoggeandH.S.Stoneproposedarecursivedou-blingalgorithmforthesolutionofageneraJclassofrecuxrenceequationsl1].Itisthefastestalgorithm(thetimeisO(log,N))whenthenumberofprocessingelemelltspiseqllaltoN.Howeveritisnotoptimalintermsofefficiency:itsspeedupisO(de),whileitsefficiencyisO(wt).TherecursivedoublingapproachcanPro…     

方兴未艾的智能网技术

智能网(IN)的概念于1984年一被提出,便在世界范围内迅速掀起了IN研究的热潮,虽然至今已有十个年头,但仍可以说,IN技术的发展才刚刚起步。本文就是基于当前国际上对IN的研究,来介绍智能网的基本概念和结构,进而阐述其设计思想及主要特征,并对IN的未来发展趋势进行了探讨,最后为我国发展IN技术提出了可行建议。     

软X射线光刻研究现状和发展

新的一代光刻设备朝着两个方向发展 ,一是趋向于大数值孔径、短波长 ;另一方面国际上正在寻求适应二十一世纪、小于 0 .1 5微米线宽的软 X射线投影光刻技术 ,预计成为下世纪制造千兆位以上超大规模集成电路的主要设备。软 X射线光刻及其应用研究是目前国际上非常活跃的高技术领域。把软 X射线光刻列为重要发展项目。本文摘要综述德国、美国、日本、俄罗斯等国近年来在软 X射线光刻关键单元技术及整机研究和发展方面的概况、预计下世纪初美国和日本将有软 X射线投影光刻机投入工艺生产线     

Categorical ML — Category-theoretic modular programming

We consider an extension of the functional programming language Standard ML with a modular structure based upon concepts in category theory such as categories, functors, natural transformations and adjunctions. In essence, we are following the categorical imperative of considering arrows as well as objects. This is intended to enforce a certain mathematical rigour on the programmer, so that the only programs that can be expressed are those with a categorical significance. The essentially algebraic nature of category theory means that we may generate equational correctness conditions for the modular structure of programs, thus separating the correctness of individual functions from that of modules. We describe this programming language, give examples of its use, and explain how it is implemented in a type system.     

高帧频电视跟踪算法的研究

在对高速运动目标进行高精度的跟踪测量时 ,常采用高帧频电视跟踪系统。针对高帧频电视象元分辨率低、处理时间短、信噪比低的特点 ,对其定位跟踪算法进行了研究。     

SFTMP——一个实现软件容错的多处理机系统

多版本软件通过设计相异性实现了软件容错。为了对这种方法进行研究，我们的课题实现了一个三版本软件系统，称之为SFTMP（SoftwareFault－TolerantMultiProcessor）。本文描述了SFTMP的硬件结构和软件执行支持环境，该环境包括同步、表决和监控功能、版本间的通讯及故障的恢复和重构。     

Symmetry properties and bifurcation analysis of a class of periodically forced chemical reactors

In this work, we discuss how periodic forcing may induce symmetry properties into mathematical models of chemical reactors. We define a class of reactors subjected to discontinuous periodic forcing, and show that all the reactors belonging to this class have spatio-temporal symmetry. This symmetry and its influence on the possible bifurcation scenarios are discussed. The bifurcation analysis is carried out with suitable discrete systems that exploit a property of the Poincaré map. In fact, it is shown that the spatio-temporal symmetry induced by the forcing makes the Poincaré map of the continuous system an iterate of another map. On this basis, a technique to implement parameter continuation methods is proposed. With such a technique, it is also possible to characterize symmetric and nonsymmetric regimes and unstable limit sets otherwise undetected with “bruteforce” approaches. Examples for reverseflow reactors and networks of n-reactors with periodically switched feed and discharge positions are presented.     

Efficient prolog programming

This paper presents some basic programming strategies for Prolog programmers. The strategies are based on knowledge about the operational behaviour of the Warren Abstract Machine (WAM). The aim is to provide simple principles for making programs both faster and less space-consuming. The programming hints given are mostly local, i.e. only a single clause or procedure need be considered at a time.     

Artificial intelligence and symbols

The introduction of massive parallelism and the renewed interest in neural networks gives a new need to evaluate the relationship of symbolic processing and artificial intelligence. The physical symbol hypothesis has encountered many difficulties coping with human concepts and common sense. Expert systems are showing more promise for the early stages of learning than for real expertise. There is a need to evaluate more fully the inherent limitations of symbol systems and the potential for programming compared with training. This can give more realistic goals for symbolic systems, particularly those based on logical foundations.