复杂网络特性对大规模多智能体协同控制的影响

随着分布式多智能体系统应用领域和系统规模的不断扩大,网络特性已成为影响系统性能的一个重要因素.通过研究和分析复杂网络特性对大规模分布式多智能体系统协同控制的影响,对多智能体系统性能的影响做出系统性分析,同时为提出大规模多智能体组织结构的优化算法提供依据.主要针对随机网络、小世界网络、网格网络和无尺度网络这4种典型复杂网络特性,从理论和仿真两方面进行分析.在理论方面,通过基于马尔可夫链的信息传输过程在不同网络结构下的建模,对比分析了信息无偏随机游走模型和智能决策模型下的传输效率.在仿真建模中,主要从智能体间信息传输效率、不同应用领域中集成协同控制效率、对网络故障恢复的影响这3个典型的多智能体系统协同控制应用对比分析复杂网络特性对系统性能的影响.研究结果表明,复杂网络特性如小世界和无尺度特性可以在相同的控制策略下形成明显的性能差异,如果设计合理的控制算法,复杂网络结构将有助于多智能体系统性能的提升.     

事件触发控制下二阶多智能体的量化一致性

针对多智能体系统网络通信过程中信息需要量化的情况,研究了二阶多智能体系统在事件触发控制下的量化一致性。基于事件触发控制策略,提出一致性协议,并采用对数量化器对控制输入进行量化处理。利用Lyapunov稳定性理论,对系统进行一致性分析,得到了多智能体系统渐近趋于一致的充分条件。仿真结果说明了理论分析的有效性。     

多速率分布式预测控制及其在热连轧活套系统中的应用

多智能体协调控制系统更适合采用分布式控制方式,但是处理智能体之间的耦合影响是分布式控制的一个难点.本文针对串联结构下的多智能体系统,提出一类多速率分布式预测控制策略,异步更新多智能体的控制律,能够充分考虑智能体之间的耦合影响,提高系统的稳定性,并给出了系统稳定的充分条件.最后,将多速率分布式控制算法应用到热连轧活套系统,仿真验证了方法的有效性和可行性.     

输入饱和约束下多智能体的有限时间一致

为解决多智能体系统在有限时间跟踪控制过程中受输入饱和影响的问题, 本文研究了输入饱和约束下的 二阶线性多智能体系统的有限时间一致和跟踪控制. 首先在无向通信拓扑下, 利用齐次函数, 设计了基于单饱和函 数的有限时间一致和跟踪控制器. 然后, 应用李雅普诺夫稳定性理论和代数图论等方法证明了控制算法的稳定性. 最后, 给出了能够使多智能体系统实现有限时间一致和跟踪的充分条件. 仿真结果验证了当系统存在输入饱和约束 时, 控制器能使多智能体在有限时间内完成跟踪任务.     

多智能体多目标跟踪算法研究

研究了多智能体同时跟踪多个目标的群集算法,采用智能体的编号对目标数求余数的方法来确定各个智能体的跟踪目标。对于势场函数,根据智能体目标是否相同建立了两种势场函数,并在此基础上确定动态系统的控制输入,使相同目标的智能体在跟踪目标的过程中形成群集,而不同目标的智能体相互分离;用李亚普诺夫稳定性理论分析了算法的收敛性。仿真结果表明,所提算法可有效解决多目标跟踪问题。     

拒绝服务攻击下领导-跟随多智能体系统的均方一致性研究

针对带有过程噪声和测量噪声的领导-跟随多智能体系统,研究拒绝服务攻击下多智能体系统的一致性问题.首先,设计基于卡尔曼滤波的状态观测器,对智能体状态进行有效准确的估计;然后,基于预测控制理论提出一种基于状态估计信息的分布式预测控制算法,从而实现领导-跟随多智能体系统的均方一致性控制,并给出拒绝服务攻击环境下实现领导-跟随多智能体系统均方一致性的充分必要条件;最后,通过数值仿真验证所提出方法的正确性和有效性.     

重放攻击下多智能体系统H∞一致性PID控制

本文针对一类带有加性噪声和乘性噪声的离散多智能体系统,研究重放攻击下多智能体系统的H∞一致性比例–积分–微分(PID)控制问题.首先,根据智能体的测量输出设计状态观测器,对智能体的状态进行有效估计,观测器设计过程中考虑了系统测量输出从传感器传输到观测器过程中受到重放网络攻击的影响.然后,利用智能体与其邻居智能体的估计状态差设计PID控制器.利用李雅普诺夫稳定性理论和代数图论,证明在该控制策略下,多智能体系统在重放攻击存在的情况下达到预期的H∞性能指标.最后,利用线性矩阵不等式(LMI)方法求解观测器和控制器增益,利用数值仿真验证了所设计的观测器和PID控制器的有效性.     

带虚拟领导者的多智能体系统的非光滑一致性

研究了在限制交互范围条件下具有一个虚拟领导者的二阶多智能体系统的一致性问题。假设多智能体系统中所有智能体均可以接收到领导者的信息，而智能体之间只有距离在一定范围内才可以进行相互通信。在相对状态反馈的线性一致性协议下，通过李雅普诺夫法，证明了该多智能体系统二阶一致性的充分条件。接着通过仿真实例验证了理论结果，并对该算法作出了总结。     

无人机飞行仿真系统多智能体建模分析与应用研究

无人机飞行仿真系统包含多个功能模块,采用多智能体(multi-Agent)技术研究无人机飞行仿真系统,分析并建立基于多智能体的飞行仿真系统的层次结构和运行机制,进而利用智能体对无人机仿真模型的各子模块模型运行过程进行建模,最后在JADE环境下建立了多智能体无人机飞行仿真系统的实例模型.这种方法可以通过仿真模型各子系统智能体之间的交互来完成无人机飞行仿真系统中飞行仿真的主要功能,从而满足无人机模拟训练系统的设计需求,在无人机模拟训练系统研究中具有广阔的应用前景.     

牵制控制下的多智能体系统群一致性

本文研究了一类多智能体系统在牵制控制下的群一致性问题,提出了融合群内信息交互、群间信息交互和牵制控制器的一致性协议.对固定拓扑下的二群组智能体系统和切换拓扑下的多群组智能体系统,利用稳定性理论和图论分别给出了适用于任意拓扑结构的充要条件,使得智能体系统在所提协议和牵制控制器的联合作用下实现预期的群一致.针对拓扑图中含有生成树这一特例,分析了被施加牵制控制的智能体在结构中的具体位置.此外,对切换拓扑下进行非线性交互的多群组智能体系统,利用Lyapunov方法推导出一充分条件,得出只要对多智能体系统的一部分主体进行牵制控制,则所有智能体即可在所提协议和牵制控制器的作用下渐近收敛于各自的期望一致平衡点.最后,仿真例子验证了所提方法的有效性.     

大飞机项目风险管理规划决策系统设计与实现

风险管理是各个工程项目必须的环节之一。本文首先对现有的风险管理体系进行介绍和分析,进而完成适用于大飞机项目的风险管理规划中三大核心要素——甘特图、概率影响矩阵和角色权限控制的设计,最后完成了大飞机项目风险管理规划决策系统的实现,为项目风险识别和分析提供了良好的数据接口。     

数字孪生丹江口工程水质安全“四预”平台研发及应用

丹江口水库的水质安全关系到南水北调中线供水的保障，数字孪生丹江口是国内首个以水利工程水质安全为业务需求之一的数字孪生工程。针对丹江口水库水体水深大、范围大、污染来源分散和类型多，以及丹库与汉库水动力交换复杂、水质呈现分区分层等特点，研发高效高精度水动力水质机理模型、水质在线动态推演技术、水质浓度场三维仿真技术，首次实现水库全域水质浓度场时空演变全过程的在线推演和三维立体仿真展示，通过海量多源异构数据的科学解析、模型集成与多业务有机融合，实现水质在线推演、预警分析、安全态势预演、预案比选决策的水质安全“四预”功能。水质安全“四预”功能已成功应用于丹江口水质安全管理，为中线水源工程供水水质保障管理提供有力支撑。     

Managing project risk using combined analytic hierarchy process and risk map

The main purpose of the study is to develop an integrated framework for managing project risks by analyzing risk across project, work package and activity levels, and developing responses.Design/methodology/approachThe study first reviews the literature of various contemporary risk management frameworks in order to identify gaps in project risk management knowledge. Then it develops a conceptual risk management framework using combined analytic hierarchy process (AHP) and risk map for managing project risks. The proposed framework has then been applied to a 1500 km oil pipeline construction project in India in order to demonstrate its effectiveness. The concerned project stakeholders were involved through focus group discussions for applying the proposed risk management framework in the project under study.FindingsThe combined AHP and risk map approach is very effective to manage project risks across project, work package and activity levels. The risk factors in project level are caused because of external forces such as business environment (e.g. customers, competitors, technological development, politics, socio-economic environment). The risk factors in work package and activity levels are operational in nature and created due to internal causes such as lack of material and labor productivity, implementation issues, team ineffectiveness, etc.Practical implicationsThe suggested model can be applied to any complex project and helps manage risk throughout the project life cycle.Originality/valueBoth business and operational risks constitute project risks. In one hand, the conventional project risk management frameworks emphasize on managing business risks and often ignore operational risks. On the other hand, the studies that deal with operational risk often do not link them with business risks. However, they need to be addressed in an integrated way as there are a few risks that affect only the specific level. Hence, this study bridges the gaps.     

基于模糊集的风险聚类预测方法

复杂社会技术系统存在许多不确定性的因素,这些因素给社会决策、项目过程管理带来了巨大的障碍和风险,因此有效的风险预测方法变得十分重要.根据风险项目的风险因素向量,利用模糊等价类的方法,对风险项目的历史数据进行模糊聚类,进而通过对新的风险项目和历史数据的模糊匹配实现了项目的风险聚类预测方法.分析和实践表明,该模型有效地解决了风险项目中诸多不确定性因素分类问题.该方法适合于政府决策、电子商务、软件项目管理等方面的风险管理应用.     

Risk of fire emergency evacuation in complex construction sites: Integration of 4D-BIM,social force modeling,and fire quantitative risk assessment

Construction industry is claimed to be the fourth most dangerous sector by number of fatalities. In complex construction sites, emergency evacuation risk assessment is a challenging task due to their ever-changing nature. This study developed a model to analyze the risk of fire emergency occurrence, and risks which are associated with evacuation performance (in response to that emergency) through an integrated approach in complex construction sites. To analyze the evacuation scenarios more realistically, we utilized Social Force Model (SFM) simulation engine. Using SFM for simulating the evacuation of complex construction sites has not been adequately addressed in the literature. Microscopically simulating the evacuation scenarios for all workdays of the studied complex project required high computation efforts. To tackle this computation challenge, a parallel computing technique was coupled with SFM simulation engine. More importantly, in this paper site’s evacuation performance was evaluated multi-objectively considering evacuation time and evacuation safety. The construction site’s emergency scenarios were modeled by 4D-BIM, potential for trigger fire emergency was determined by a fire ignition Quantitative Risk Assessment (QRA) module, and site evacuation was simulated by SFM simulation engine. The proposed framework handled the collaboration of 4D-BIM, fire QRA module, and SFM engine. This research study benefited from data driven from a real mega project. The findings demonstrated that analyzing the risk of evacuation through an integrated approach by the proposed model could render more realistic results. The results also provided the project managers with a reliable safety decision-making support.     

支持HLA仿真和并行绘制的统一对象模型研究

随着计算机硬件、软件、网络以及社会需求的发展,分布式交互仿真应用对大规模复杂场景和大量仿真实体对象的管理能力和绘制能力提出了较高的要求.针对现有HLA仿真应用中大规模复杂场景的视景仿真实时性较差、仿真实体对象管理与绘制对象管理是分离的等问题,提出了统一对象模型,包含异质实体对象树、操作记录列表和统一访问接口,实现了仿真实体对象和绘制对象的高效组织和统一管理.该统一对象模型在HLA仿真平台和并行绘制平台之间建立了高效的数据交换桥梁,减轻了两者集成的开发工作量,并有效地支持了大规模复杂场景和大量仿真实体对象的实时绘制,有助于提高仿真的实时性.该统一对象模型对于HLA仿真平台和并行绘制平台具有通用性.最后给出了实验结果和分析.     

An information architecture for risk assessment and management

As systems become more complex, project managers need a means of accessing risk management experience gained on other projects. To address this need, the authors developed RAMP (Risk Assessment and Management Program), a risk management information system that provides interactive support for identifying, analyzing and sharing risk mitigation experience     

Computers in engineering risk and hazard management

Summary The engineering of large scale facilities, such as dams, power stations, bridges etc, involves the handling of large amounts of information. Managers of the design and construction process have to take on a wide range of roles to cope with it all. One important aspect of this information is that concerned with safety, risk and hazard management. This paper is divided into three sections each covering different aspects of a common approach to this problem. The analysis of risk using traditional reliability techniques is not covered. The concern here is rather with the use of computers to support and inform the direct management of quality, safety and hazard and hence to indirectly control risk. Firstly, the approach based on the use of “Interacting Objects” will be outlined. This will be illustrated through the use of IT to support business processes in quality management. Product and process models will be compared. Safety, risk and hazard are part of quality. Secondly, the use of these objects in physical process simulation will be described. Here the motivation for the work is to begin to look at the implications for risk analysis of the sensitivity of the behaviour of simulated non-linear systems to initial conditions. Thirdly, the identification and management of “proneness to failure” in a project will be outlined. Here the problem is how to deal with the difficult interaction between technology and human and organisational factors.     

Knowledge-based risk assessment and cost estimation

A knowledge-based method for software project risk assessment and cost estimation has been implemented on multiple platforms. As an extension to the Constructive Cost Model (COCOMO), it aids in project planning by identifying, categorizing, quantifying and prioritizing project risks. It also detects cost estimate input anomalies and provides risk control advice in addition to conventional COCOMO cost and schedule calculation.The method has been developed in conjunction with a system dynamics model of the software development process, and serves as an intelligent front end to the simulation model. It extends previous research in the knowledge-based cost estimation domain by focusing on risk assessment, incorporating substantially more rules, going beyond standard COCOMO, performing quantitative validation, providing a user-friendly interface, and integrating it with a dynamic simulation model.Results of the validation are promising, and the method is being used at Litton Data Systems and other industrial environments. It will be undergoing further enhancement as part of an integrated capability for software engineering to assist in system acquisition, project planning and risk management.