Multi Agent动态分组模型用于多无人机协同任务分配

多无人机协同任务分配应用Multi Agent系统模型时需考虑对大规模Agent的分布式管理和动态协调。协同任务分配将每一无人机视为系统中一个Agent,对全体Agent进行动态分组,分别提出决策Agent及成员Agent的动态分组模型和评价函数,由决策Agent管理和协调本组全体Agent执行任务,降低了大规模任务分配的决策成本并有利于实现全体Agent协同。仿真实验验证了模型及算法有效性。

Multi Agent在分布式测控系统动态任务调度中的实现

Multi Agent实现是基于Multi Agent的分布式测控系统动态任务调度算法实现的关键技术。采用Java作为开发工具,根据Multi Agent的功能,详细论述了Multi Agent在分布式测控系统动态任务调度中的实现。基于Multi Agent的动态任务调度算法根据各主机的负载状态,在系统运行过程中利用移动Agent动态迁移任务。文中详细论述了利用Aglets系统开发和执行移动Agent,从而有效地提高了系统效率,实现了动态任务调度的目标。

基于Multi Agent的图象理解

最近十多年来 ,具有专家系统外壳的图象理解软件的开发一直是一个重要的研究课题。然而 ,多数基于知识的图象理解系统或者局限于诸如分割之类的特定图象操作上 ,或者局限于像理解建筑物的航空图象这样特定的应用。由于这些系统的目标和知识结构的局限性 ,因而不能将它们推广到其它领域。此研究就是试图解决此问题 :首先将Agent构造环境扩展为一个MultiAgent的图象理解系统 ,而此系统是知识入口的开发工具 ,它为用户提供一个类似于专家系统外壳的界面。这样一方面能加速图象理解系统的开发 ,另一方面方便那些缺乏图象理解知识的人开发图象理解应用程序。

一种基于Multi Agent的无线传感器网络数据收集算法

提出了构造多Agent协作模型的方法,设计了基于多Agent的无线传感器网络的数据收集算法ARDG(activereliable data gathering algorithm),将多Agent的协作思想结合到数据路由的计算中,该算法在无线传感器网络资源限制,如在传感器存储空间和能量有限等条件下,提高了数据收集的主动性和可靠性,理论和仿真实验证明了该方法的可靠性与时效性。

Modelling urban expansion using a multi agentbased model in the city of Changsha

Although traditional urban expansion simulation models can simulate dynamic features, these models fail to address complex changes produced by different agents＇ behaviors. The paper has built up a set of spatial-temporal land resource allocation rules and developed a dynamic urban expansion model based on a multi-agent system, which can simulate the interaction among different agents, such as residents, peasants, and governments. This model is applied to simulate urban expansion process taking Changsha City, in China as a study area. The results show that this model can not only reflect basic characteristics of urban expansion, but also help explain the reasons for urban expansion process and understand the effect of agents＇ behavior on the expansion process, and provide insights into the causing factors behind the expansion. In addition, in contrast to simulation results with land use classification map from remote sensing images, the precision of the simulation reached over 68% with higher precision than cellular automata model according to the cell-by-cell comparison. The results suggest that the model can help to provide land use decision making support to government and urban planners.

Towards a Comprehensive Security Framework of Cloud Data Storage Based on Multi Agent System Architecture

The tremendous growth of the cloud computing environments requires new architecture for security services. Cloud computing is the utilization of many servers/data centers or cloud data storages (CDSs) housed in many different locations and interconnected by high speed networks. CDS, like any other emerging technology, is experiencing growing pains. It is immature, it is fragmented and it lacks standardization. Although security issues are delaying its fast adoption, cloud computing is an unstoppable force and we need to provide security mechanisms to ensure its secure adoption. In this paper a comprehensive security framework based on Multi-Agent System (MAS) architecture for CDS to facilitate confidentiality, correctness assurance, availability and integrity of users' data in the cloud is proposed. Our security framework consists of two main layers as agent layer and CDS layer. Our propose MAS architecture includes main five types of agents: Cloud Service Provider Agent (CSPA), Cloud Data Confidentiality Agent (CDConA), Cloud Data Correctness Agent (CDCorA), Cloud Data Availability Agent (CDAA) and Cloud Data Integrity Agent (CDIA). In order to verify our proposed security framework based on MAS architecture, pilot study is conducted using a questionnaire survey. Rasch Methodology is used to analyze the pilot data. Item reliability is found to be poor and a few respondents and items are identified as misfits with distorted measurements. As a result, some problematic questions are revised and some predictably easy questions are excluded from the questionnaire. A prototype of the system is implemented using Java. To simulate the agents, oracle database packages and triggers are used to implement agent functions and oracle jobs are utilized to create agents.

Cement industry control system based on multi agent

Cement production is characterized by its great capacity, long-time delay, multi variables, difficult measurement and multi disturbances. According to the distributed intelligent control strategy based on the multi (agent,) the multi agent control system of cement production is built, which includes integrated optimal control and diagnosis control. The distributed and multiple level structure of multi agent system for the cement control is studied. The optimal agent is in the distributed state, which aims at the partial process of the cement production, and forms the optimal layer. The diagnosis agent located on the diagnosis layer is the diagnosis unit which aims at the whole process of the cement production, and the central management unit of the system. The system cooperation is realized by the communication among optimal agents and diagnosis agent. The architecture of the optimal agent and the diagnosis agent are designed. The detailed functions of the optimal agent and the diagnosis agent are analyzed. At last the realization methods of the agents are given, and the application of the multi agent control system is presented. The multi agent system has been successfully applied to the off-line control of one cement plant with capacity of 5 000 t/d. The results show that the average yield of the clinker increases 9.3% and the coal consumption decreases 7.5 kg/t.

Multi Agent Based Power Distribution System Restoration—A Literature Survey

Multi agent system (MAS) is one of the most dominant research wings which consist of several agents who interact with each other to achieve a common objective. MAS has been developed for a wide range of applications in power systems. Power system restoration is a main application of that. Researchers present several architectures for fault identification, isolation and restoration of the power system. This paper presents a complete literature review on available architectures for power distribution restoration and future trends in MAS based power system restoration.

Dual Layered Architecture for Multi Agent Based Islanding and Load Management for Microgrids

This paper proposes a novel dual layered multi agent system (MAS) based control system for the use in microgrid operations. In developing a smarter grid capable of withstanding disturbances and/or outages and providing quality service to the consumers, reliable microgrid control architecture is vital. The innovative microgrid control system proposed, makes the microgrid capable of isolating the local grid from effects of any upstream disturbances in the main utility grid by operating disconnected from the main utility via islanding, and it allows the most critical local loads to be supplied by any, available, local power source during such islanded operation. The proposed MAS control architecture is developed using the JADE platform and it is used to control a test network simulated in MATLAB. The results of these simulations show the capability of developing MAS based reliable control mechanism for islanding and load management of microgrids based on the proposed concept.

A New Formalism for Modeling a Multi Agent Systems: Agent Petri Nets

In this paper, we present a new formalism for Modeling Multi Agent Systems (MAS). Our model based a PN is able to describe not only not the internal state of each agent modeled but also its behavior. Owing to these features, one can model naturally the dynamic behavior of complex systems and the communication between these entities. For this, we propose mathematical definitions attached to firing transitions. To validate our contribution, we will deal with real examples.

Multi Agent Knowledge Management Architecture

Nowadays, knowledge in Public Sector environment becomes very vast and increasing day by day at speedy pace. So, to handle and manage the knowledge becomes a tedious job, resulting into degrading the overall affectivity and productivity of the system. Hence, the need of effective architecture arises, which can increase the performance of disseminating knowledge in public sector. This results the implementation of knowledge management (KM) using Multi Agents (MA). Using Multi Agents reduces the time overhead for serving relevant knowledge to end users. The objective of this paper is to propose KM architecture using MA which will be helpful and effective in circulating knowledge to public sectors in a much better and easier manner, due to which it enhances the productivity and performance. The paper firstly, gives the understanding of literature on various knowledge management frameworks and tools for implementing Multi Agents. Then it proposes a MA enterprise knowledge management architecture (MAEKM), stating that how knowledge circulation will be done. At the end, using JADE framework, paper implements MAEKM architecture for public sector. The paper describes the necessity of implementing this architecture and its usefulness in disseminating knowledge in public sectors.

基于Multi-Agent的无人机集群体系自主作战系统设计

针对无人集群自主作战体系设计中的关键问题,提出基于Multi-Agent的无人集群自主作战系统设计方法。建立无人集群各节点的Agent模型及其推演规则;对于仿真系统模块化和通用化的需求,设计系统互操作式接口和无人集群自主作战的交互关系;开展无人集群系统仿真推演验证。仿真结果表明,所提设计方案不仅能够有效开展并完成自主作战网络生成-集群演化-效能评估的全过程动态演示验证,而且能够通过重复随机试验进一步评估无人集群的协同作战效能,最后总结了集群协同作战的策略和经验。

Multi-Agent Deep Reinforcement Learning for Cross-Layer Scheduling in Mobile Ad-Hoc Networks

Due to the fading characteristics of wireless channels and the burstiness of data traffic,how to deal with congestion in Ad-hoc networks with effective algorithms is still open and challenging.In this paper,we focus on enabling congestion control to minimize network transmission delays through flexible power control.To effectively solve the congestion problem,we propose a distributed cross-layer scheduling algorithm,which is empowered by graph-based multi-agent deep reinforcement learning.The transmit power is adaptively adjusted in real-time by our algorithm based only on local information(i.e.,channel state information and queue length)and local communication(i.e.,information exchanged with neighbors).Moreover,the training complexity of the algorithm is low due to the regional cooperation based on the graph attention network.In the evaluation,we show that our algorithm can reduce the transmission delay of data flow under severe signal interference and drastically changing channel states,and demonstrate the adaptability and stability in different topologies.The method is general and can be extended to various types of topologies.

基于多Agent传动关系的股市趋势预测

股市趋势预测是机器学习领域中一个具有挑战性的任务。由于一些因素对于股市的影响是动态且不确定的,导致股市趋势难以预测。针对已有方法在股市预测时存在的灵敏性差、适应力弱等问题,从快变量和慢变量的传动关系出发,利用Agent技术对股市中的快周期和慢周期进行联合建模,提出一种多Agent传动影响图(MATID)股市趋势预测方法。给出股市中快周期和慢周期的划分标准,并引入周期能量的概念;通过对相关趋势指标的特征融合,给出周期能量的量化计算方法;通过分析快周期和慢周期的动态作用过程,给出传动因子的表示方法;将快周期和慢周期分别对应成不同的Agent,利用多Agent影响图模型建模快周期和慢周期的传动过程;利用股市振子模型表示快Agent和慢Agent之间的传动效用,利用联合树的自动推理技术对股市趋势进行预测。在不同样本数量和不同股市趋势下进行实验,结果表明,与门控循环单元、S-LSTM和Hybrid-RNN预测方法相比,MATID方法预测精确率提升1.5%~7.0%,召回率提升5.4%~6.7%,F1值提升3.7%~6.2%,具有良好的灵敏性和适应力。

竞争与合作视角下的多Agent强化学习研究进展

随着深度学习和强化学习研究取得长足的进展,多Agent强化学习已成为解决大规模复杂序贯决策问题的通用方法。为了推动该领域的发展,从竞争与合作的视角收集并总结近期相关的研究成果。该文介绍单Agent强化学习;分别介绍多Agent强化学习的基本理论框架——马尔可夫博弈以及扩展式博弈,并重点阐述了其在竞争、合作和混合三种场景下经典算法及其近期研究进展;讨论多Agent强化学习面临的核心挑战——环境的不稳定性,并通过一个例子对其解决思路进行总结与展望。

基于Multi-Agent技术的林产品供应链竞合研究

林产品贸易快速发展,随之而来的是竞争日益加剧,林产品供应链管理是林业产业发展的必然结果,供应链上企业间的竞合研究成为了热点。为此,搭建了基于Multi-Agent的林产品供应链,并在探讨核心制造商和销售商博弈收益的基础上,针对供应链中两者竞合关系进行了研究。通过对相应情况的建模仿真,在实验结果对比分析中发现制造商和销售商合作关系的影响因素有:初始合作比例、契约违约金和第三方监督,三者对合作行为有不同的影响,高水平的惩罚和监督可以有效抑制合作过程中的背叛行为。在此基础上,提出了培养信任文化、建立信誉评价平台和增设第三方监督主体的促进供应链合作的建议,为林产品供应链竞合研究提供参考。

基于多Agent的锂电池主动均衡策略控制仿真研究

【目的】针对锂电池的荷电状态均衡管理问题,提出一种基于多智能体的电池组荷电状态一致性均衡方案。【方法】首先,将多智能体控制策略引入电池管理的下垂控制中,实现了主动均衡电路拓扑下的自主均衡;其次,建立领航跟随者模型,利用参数已知的虚拟智能体使各个荷电状态不一致的电池的状态向其靠近,实现充放电模式下的荷电状态均衡;最后,对二阶多智能体荷电状态均衡控制策略进行仿真验证。【结果】实验结果表明,相比一阶均衡控制策略,自主均衡时间减少了43.02%,充电模式中均衡时间减少了16.13%,放电模式中均衡时间降低了32.90%。【结论】多智能体系统在电池的均衡管理中能够实现荷电状态的均衡,有效地降低了锂电池荷电状态到达一致性的收敛时间。

基于组合遗传算法的电力科技创新成果多agent集成仿真

电力科技成果普遍交叉,为了降低集成电力科技创新成果耗时与能量损耗,研究基于组合遗传算法的电力科技创新成果多agent集成仿真。将底层数据库中电力科技创新资源经多种接口调配至关联库、综合评价以及专家评价三大agent模块,通过结合遗传算法与蚁群算法优点的组合遗传算法,优化解决多agent集成任务分配问题,实现电力科技创新成果高效集成。实验表明,该算法的寻优能力高、求解速度快;可规范化、结构化地集成电力科技创新成果信息;具有耗时少、能量损耗低优势。为科技创新和成果转化提供新思路。