多Agent系统安全性分析与设计

全面分析了多Agent系统的安全特性、安全风险和安全需求.在此基础上,研究了多Agent系统多层安全模型及安全多Agent系统需要提供的基础安全服务.基于FIPA-OS框架,设计并实现了一个面向Agent的二级认证PKI系统,同时提出一组安全服务Agent,为FIPA-OS框架增加了的身份鉴别、证书管理、安全审计、资源访问控制及Agent社区管理等多种安全机制.     

多Agent系统安全模式设计

设计了一种使用SPKI证书体系结构为基础的多Agent安全模式结构，保证了单个Agent活动的可验证性和不可抵赖性，并由此构建了一个以此安全铆为对象的多Agent系统安全模式结构。     

一种基于Agent技术的安全能力自动协商机制

适应多安全域环境的安全能力自动协商机制包括了一种基于Agent的安全能力协商模型、适用于该模型的协商Agent和协商流程。该机制中，通过定义安全能力协商描述语言(SanDL)文档，用户可以方便准确地将协商要求和协商策略告知协商Agent，由协商Agent自动完成协商过程。在协商Agent中通过定义通信适配器和一套基本协商原语，使协商Agent的通信接口更加安全、灵活、易于扩展。     

多Agent专家系统安全性设计与研究

针对多Agent专家系统面临的安全威胁,提出了单个Agent的安全体系结构及多Agent专家系统安全架构模型。利用容器砂箱模型,实现了多Agent专家系统基于角色的访问控制。讨论了PKI在多Agent专家系统中的实际应用,在此基础上,研究了多Agent专家系统身份验证、安全通信、安全管理、安全审计及安全恢复机制。应用开发实践表明,系统具有资源占用少、安全性高、稳定性好以及可伸缩性强等优点,可应用于企业网智能安全管理、电子商务等领域。     

基于门限技术的Mobile Agent系统安全性设计

针对移动代理的安全问题，提出了基于门限技术的多Agent系统安全性设计方案。方案采用门限秘密共享技术，将加密信息和密钥由多个移动Agent分开携带，保护移动Agent的隐私，提高移动Agent系统的可靠性和容错能力：通过有限授权实现移动Agent代表用户在远端签约而不暴露其主人的私钥的功能。     

一种基于Multi-Agent的实时监控系统的设计和实现

基于Multi-Agent技术提出了一种多Agent实时安全支持系统结构模型，包括：故障感知Agent、故障诊断Agent和咨询解释Agent。它们的协同工作用于某超高压变电站安全监控环境、为大型实时监控系统的开发提供了一个成功的实例。     

对FIPA多Agent系统的攻击行为分析

本文通过对JADE多Agent系统的攻击分析，研究了目前基于FIPA规范的多Agent系统的安全弱点，并提出了一个采用身份证书及授权策略来加强多Agent系统安全的框架。     

多Agent系统DOLTPI一Agent中的通信及其安全机制

本文介绍了一个基于Internet的信息检索多Agent系统DOLTRI-Agent内多个Agent间的通信框架及其安全机制.主要论述各个Agent间相互通信所采用的通信语言,通信协议以及信息交换和相互合作的方法,在此基础上提出了一种多Agent系统的安全通信模型.     

基于MAS的复杂系统分布式求解策略与推理研究

针对复杂的连续生产线系统监测与故障诊断的动态、分布、实时和不确定特性，将多Agent及相关智能技术引入复杂故障诊断领域，提出了一种嵌套式基于消息传递的多Agent组织模型，分析了基于MAS的分布式智能故障诊断方法和过程；研究了模型系统的动态适应性和稳定性；设计了实时诊断Agent工作状态的表达机制；讨论了Agent间的协调协作机制及融合方法；给出了多Agent诊断系统诊断决策的集成描述结构；提出了任务分解与分配调度算法，以充分发挥Agent的社会性和基于场景的特点，使得在复杂系统的问题求解中能提供更为可靠的诊断结果，而且还能节约资源提高诊断效率．将其应用于某安全监控系统中，取得了与专家相似的诊断结果，克服了以往监控诊断系统存在的弊端，提高了企业的安全运行效率．与传统诊断方法相比，体现了Agent技术在复杂分布式问题求解领域的特有优势和良好前景．     

移动Agent的安全认证方案

移动Agent为分布式计算提供了一种新的处理模式，但是安全问题阻碍了它的发展。本文简要介绍了几种常用的安全认证技术，然后针对移动Agent的安全问题，根据移动Agent和Kerberos协议的特点。将Kerberos协议加以改进并应用于移动Agent的身份认证系统，提出了一种新的安全认证机制，提高了移动Agent在Intemet等复杂应用环境下的安全等级。     

基于Modelica的虚拟远程实验室

为了弥补传统实验教学模式的不足,介绍一种使用新型的、面向对象的多领域统一建模语言Modelica进行建模及仿真的电工电子虚拟远程实验室,并详述了其架构及远程实验的操作流程。构建两个子系统——模电实验系统及吹摆控制实验系统。通过调用Modelica标准模型库,实现了模电实验系统的建模;根据模块化建模的思想,对吹摆控制实验系统进行系统分解,利用Modelica编写每个模块的仿真模型,建立了可扩展的系统模型库,并基于该模型库实现了吹摆系统的构建。将两个子系统放在虚拟实验平台上仿真,仿真结果符合理论推算。     

Methods of Continuous Logic in Control

Continuous logic, its problems, and methods were outlined, and its basic operations were defined. The algebra of continuous logic was described, and its main functions of one, two, and three variables were listed. The laws of this logic were presented and contrasted with the laws of the discrete two-valued logic. Described were the problems of listing all continuous-logic functions of a given number of variables and representing them in a standard form. The difference between these forms and their counterparts in the two-valued logic was shown. Minimization procedures for the continuous-logic functions and their decomposition into functions of a smaller number of variables were described. The distinctions of these procedures from their counterparts in the two-valued logic were noted. The problems of analysis and synthesis of the continuous-logic functions were formulated, and methods for their solution were presented. The problem of synthesis was shown not to be necessarily solvable. The fundamentals of the continuous-logic differential and integral calculuses were presented. Any continuous-logic function was shown to have no-derivative points. The problem of completeness was described for the continuous logic together with the existing results and their distinctions from the discrete case. Numerous applications of the continuous logic to mathematics, engineering, economics, social sciences, and so on were described, and its perspectives were estimated.     

基于混合神经网络的智能问答算法

为让用户通过输入自然语言就可以跟机器进行交互,实现文本的智能问答,提出基于混合神经网络的智能问答算法。将LSTM(long short-term memory)和CNN(convolutional neural network)相结合。利用LSTM计算问题和答案的语义特征,针对语义特征的选择进行改进。采用CNN对LSTM得到的语义特征进行筛选;通过计算问题和答案特征之间的相似度得到该模型的目标函数,给出问题对应的正确答案。仿真结果验证了该算法的可行性及有效性。     

Work-related and non-work-related musculoskeletal symptoms

The purpose of the study was to analyse the prevalence of self-reported work-related and non-work-related musculoskeletal symptoms among white- and blue-collar workers. The study was carried out at eight metal industry companies and included 450 subjects, where 241 were blue-collar workers and 209 were white-collar workers. Among the blue-collar workers 39% were females and among the white-collar workers 35% were females. The study was based on a questionnaire, using the general standardized Nordic Musculoskeletal Questionnaire (NMQ). The NMQ was completed with a question concerning whether or not the symptoms were believed to be related to the individual's present work. It was found that the differences between white- and blue-collar workers' musculoskeletal symptoms became greater when solely work-related symptoms were included, compared with symptoms in general according to the NMQ. Furthermore, it was found in both groups that the prevalence of musculoskeletal symptoms decreased substantially when solely work-related symptoms were included, compared with symptoms in general.     

基于智能网格节点的计算包特征提取分析

介绍了网格的概念，通过对数据网格发展历程的分析阐明了对数据网格智能化发展方向的一些观点。分析了节点处计算包的特征形成，给出了计算包的特征提取算法，并将贝叶斯风险决策模型运用到失效计算包的剔除中。论述了计算包的特征分析对数据网格系统智能化的现实意义，并指出了实现数据网格智能化要注意的若干问题，最后对智能化数据网格系统的应用前景做了进一步的展望。     

体育成绩测试中BP神经网络的应用分析

基于MATLAB建立了一种BP神经网络算法模型,介绍了其计算流程和计算代码。该BP神经网络模型在运动员3000 m长跑成绩仿真测试中表现出了较高的准确度和可信度,5次实验的平均误差仅为0.18 min,最小误差仅为0.1 min。还利用该BP神经网络模型研究了运动晨脉、血压、血氧与体育成绩之间的关系。其中晨脉和血压稳定状态下,运动员的成绩较为稳定,且出现提升趋势;其收缩压和舒张压均处于临界值时,运动员的状态较为兴奋,相应成绩较好。     

基于广义包围盒的交互操作在Vega环境中的应用

介绍了视景仿真软件Vega所具备的场景基本操作功能，针对Vega在交互操作功能方面的不足，提出了一种广义包围盒。介绍了广义包围盒的实现原理，定义了包围盒的数据结构，给出了获取方法，然后基于Vega API，通过两种不同方法建立了场景模型广义包围盒。基于广义包围盒，对场景模型直接进行操作，从而实现了Vega虚拟场景中的模型拾取、碰撞检测等仿真交互操作。给出了模型拾取、碰撞检测的具体实现算法，并成功应用在一个视景仿真程序中。     

Evaluation of a finite element modelling approach for welded aluminium structures

Experimental and numerical studies were carried out to investigate the behaviour of quasi-statically loaded fillet-welded connections in aluminium alloy EN AW 6082. The components were designed so that fracture should occur in the heat-affected-zone (HAZ), which was also the case in the experiments. In the numerical study, the components were modelled in LS-DYNA using shell elements. A user-defined material model comprising the Barlat and Lian anisotropic yield criterion and a ductile fracture criterion was adopted. The strength and hardening data for the HAZ, weld and base material were taken from existing experimental data in the literature. The constants for the yield criterion were identified from uniaxial tensile tests by various methods, and the set of constants that best represents the measurements was adopted in the numerical analysis. Reasonable estimates on ductility were obtained by the simulations when a refined mesh was used, while the strength was somewhat over-predicted.     

Analysis of EEG tracings in frequency and time domain in hepatic encephalopathy

Spectral EEG analysis in hepatic encephalopathy (HE) is usually performed disregarding the effect of epoch length, statistical errors and equipment noise. A study on these items was carried out. In addition, spectral analysis and a new analysis, performed in time domain, were compared in the assessment of HE. The EEG tracings of 73 cirrhotic patients with HE were analyzed. Artifact-free periods of about 1 min were selected. Equipment noise was measured by short-circuiting all the electrodes. The equipment noise was notable below 1.5 Hz; the best epoch length was 4s and the statistical errors were minimal for the band with the highest relative power. Nineteen per cent of the tracings were unstable. The spectral values were found to be related to liver function and to the degree of HE, whereas the relationship with psychometric variables was poor. The indexes computed by time-domain analysis were found to be better related to psychometric findings. We have provided information on the optimisation of spectral EEG analysis and presented a time-domain analysis giving results related to psychometric tests and liver function.     

3D reconstruction based on hard tissue microtome cross-section pictures in dentistry

The aim of this study was to examine the accuracy of the computerized 3D surface analyzing and volume measuring method in dentistry. Two different types of test objects were used in the first part of the measurements. Each sample of the two groups was cross-sectioned using a hard tissue microtome. The sections were photographed on both sides and were projected on a graphical tablet and analyzed using a computer program. The measured and calculated parameters were compared. In the second part, 200 microm thick horizontal sections were prepared from 11 human incisor roots using the hard tissue microtome. This way, five sections were prepared from the apical 2 mm of each root. The effects of section thickness and number were modeled by decreasing the inclusion rate of the obtained number of sections from 10 to 2 and its influence on the calculated results was determined. This method was suitable for the approximation and analysis of 3D parameters. The results indicated that using 200-300 microm section thickness, the measured values were approximately 8-21% lower than the real parameters.