一种面向Agent的概念化建模方法*

面向Agent的建模方法为基于Agent系统的概念化描述提供了一种新的技术.针对多Agent系统分析与建模问题,扩展并规范了面向Agent概念化建模过程,定义了Agent模型及其扩展模型,并给出了形式化描述.同时,给出面向Agent的分析和设计过程,为其具体建模提供了一种高层指导.最后,与其它现有面向Agent的方法在应用方面进行了类比分析,为该方法的广泛和深入地应用研究打下基础.     

面向态势估计的多代理规划识别模型的研究

态势估计问题是数据融合的高层部分,如何有效的评估态势进而推理出敌军的目的、意图是要解决的问题;而在战役对象多、机动复杂的情况下如何推理更是一个待解决的问题。通过应用多代理规划识别理论,建立态势评估的推理框架,进而推出解决态势融合的推理算法。     

连续信息空间中的信息泄漏问题研究

在联邦式作战仿真平台中,互相通信的主体在进行通信的同时还需要保护其私有的信息。讨论了基于线性方程组“无穷解”条件下的连续信息空间的信息泄漏和保护问题,包括该模型的基本特性和防止泄漏的方法,指出了存在信息的部分泄漏问题,并给出了检测部分泄漏的方法,为提高计算性能,可指定检测子空间,最后,讨论了信息动态改变对信息泄漏的影响。     

新型潜伏性环氧树脂体系固化动力学

采用非等温差示扫描量热(DSC)技术对改性咪唑类固化剂(MIM)及其微胶囊固化剂(MIC)与E-51环氧树脂的固化反应过程进行了跟踪,并利用Kissinger和Crane方程对该固化反应进行了动力学分析,在此基础上探讨了固化剂包覆处理前后其环氧树脂体系的固化动力学参数与固化剂室温贮存性能的关系。结果表明:在不同升温速率下,E-51/MIM反应体系的放热量均大于E-51/MIC体系;两固化反应体系的反应级数均为0.89;与E-51/MIM体系相比,E-51/MIC体系的固化反应活化能和频率熵因子均较大,并具有更好的室温贮存性能。     

一种面向移动Agent的多任务并行计算模型及算法

介绍了一种面向移动Agent的并行计算模型,给出了采用十标度策略解决任务排序,采用满射策略解决任务映射的算法。该模型允许多个计算任务在异构主机构成的分布式环境下同时进行计算,并且通过算法优化,降低移动Agent之间的通信成本,减少网络流量。     

军事网格与Agent Grid研究综述

在简要介绍通用网格概念和发展的基础上,着重对军事网格和Agent Grid进行了研究和分析.介绍了美军军事网格发展历程中4个具有代表性的网格研究项目,展现了当前军事网格研究的现状和未来的发展趋势.以CoABS Grid项目为研究背景,分析了Agent Grid的基本特点和相关的主要观点,着重探讨了Agent和Grid技术的融合,以及Agent Grid与MAS的关系,最后总结提出了Agent Grid研究的技术要点.     

基于行为建模的移动代理性能评估

针对移动代理的分布式应用性能评估需求,建立了移动代理生命周期中主要行为的网络负载模型和时间消耗模型.在3种典型分布式应用想定下,分析了基于C＼S模式和移动代理模式的应用实现基本过程,对两种模式下的网络负载和执行时间进行了仿真比较,结果证明了在访问节点数量较多,网络传输速率较低时,移动代理模式的具有更好性能.     

Patents and Dual-use Technology: An Empirical Study of the World's Largest Defence Companies

Abstract

This paper examines the generation of technological knowledge by leading companies in the defence industry. In particular, we test whether the characteristics of large defence companies are related to both the production of different types of patents (civilian, military and mixed), and the generation of dual-use technologies. To explore these links, we rely on economic data for the top 100 defence companies from the Stockholm International Peace Research Institute database, and patent information from the Worldwide Patent Statistical Database. Our results show that the relationship between the production of civilian patents and the size of the company is positive and significant. However, this relationship does not hold for the production of military patents. Furthermore, the military commercial profile is unrelated to the generation of military patents. Regarding the involvement in dual-use technologies, firms engaged in dual-use are those with higher military sales, a greater number of employees and a larger number of patents (civilian, military and mixed) than those not engaged in dual-use. Furthermore, we found a skill effect (more involvement in dual-use per employee) in European firms compared to US firms. These findings help to identify which firms should be targeted by government policies if increasing dual-use technologies becomes a political objective.     

网络中基于Agent的信息搜索特点及策略规划

基于网络条件下信息搜索的特点,采用搜索策略表示信息收集Agent对主机进行信息查寻的规划和过程,将搜索过程看作是一个三维空间,这个空间由条件、方法、信息源构成;给出了一种基于Agent的信息查找策略,用AgentSpeak(L)描述并实现了规划过程.     

The impact of operational decisions on the design of salesforce incentives

When facing high levels of overstock inventories, firms often push their salesforce to work harder than usual to attract more demand, and one way to achieve that is to offer attractive incentives. However, most research on the optimal design of salesforce incentives ignores this dependency and assumes that operational decisions of production/inventory management are separable from design of salesforce incentives. We investigate this dependency in the problem of joint salesforce incentive design and inventory/production control. We develop a dynamic Principal‐Agent model with both Moral Hazard and Adverse Selection in which the principal is strategic and risk‐neutral but the agent is myopic and risk‐averse. We find the optimal joint incentive design and inventory control strategy, and demonstrate the impact of operational decisions on the design of a compensation package. The optimal strategy is characterized by a menu of inventory‐dependent salesforce compensation contracts. We show that the optimal compensation package depends highly on the operational decisions; when inventory levels are high, (a) the firm offers a more attractive contract and (b) the contract is effective in inducing the salesforce to work harder than usual. In contrast, when inventory levels are low, the firm can offer a less attractive compensation package, but still expect the salesforce to work hard enough. In addition, we show that although the inventory/production management and the design of salesforce compensation package are highly correlated, information acquisition through contract design allows the firm to implement traditional inventory control policies: a market‐based state‐dependent policy (with a constant base‐stock level when the inventory is low) that makes use of the extracted market condition from the agent is optimal. This work appears to be the first article on operations that addresses the important interplay between inventory/production control and salesforce compensation decisions in a dynamic setting. Our findings shed light on the effective integration of these two significant aspects for the successful operation of a firm. © 2014 Wiley Periodicals, Inc. Naval Research Logistics 61: 320–340, 2014