物联网中基于Multi-Agent的物间动态关系发现

物联网中存在大量异构关系的实体，其信息间的交互造成了物联网的内在矛盾。针对这一问题，提出将物联网中实体抽象化为对应的Agent，并以个体人为中心，利用本体的语义匹配及改进的物间动态关系计算方法，计算出针对个体人的物间动态关系；将原本异构的实体关系转化为Agent间基于动态关系紧密度排序的网络关系。经实验验证，该方法构建的关系网络可明显改善异构实体间交互的矛盾，而且相比原有类似方法，该方法解决问题的准确率和误差率均有改善，说明了该方法的准确性和可行性。     

When Does Communication Learning Need Hierarchical Multi-Agent Deep Reinforcement Learning

Abstract

Multi-agent systems need to communicate to coordinate a shared task. We show that a recurrent neural network (RNN) can learn a communication protocol for coordination, even if the actions to coordinate are performed steps after the communication phase. We show that a separation of tasks with different temporal scale is necessary for successful learning. We contribute a hierarchical deep reinforcement learning model for multi-agent systems that separates the communication and coordination task from the action picking through a hierarchical policy. We further on show, that a separation of concerns in communication is beneficial but not necessary. As a testbed, we propose the Dungeon Lever Game and we extend the Differentiable Inter-Agent Learning (DIAL) framework. We present and compare results from different model variations on the Dungeon Lever Game.     

硅烷自交联聚乙烯的制备与性能研究

通过熔融共混将产水剂加入到硅烷交联聚乙烯中,制备出室温下可进行自交联的硅烷交联聚乙烯材料。采用热重分析仪、平板流变仪、万能试验机对硅烷自交联聚乙烯的结构、性能进行表征。结果表明,产水剂的加入可以在自然放置的条件下逐步提高材料的凝胶含量,最终达到水煮工艺处理同样的效果;乙酸锌的促进效果最好,兼具高凝胶含量的同时,使材料的力学性能、热稳定性、储能模量提高。     

Investigation on concentration characteristics of ozone micro-bubbles fixed in ice and ozone gas released from ice

Ice formed from water in which ozone gas is dissolved is promising for the cold storage of foods because of the ozone's sterilization and deodorization capabilities. However, effective dissolution of ozone gas in water and taking ozone gas into ice are not easy. Furthermore, the decomposition rate of the ozone itself is usually very fast, regardless of its phase. Thus, to effectively take ozone gas into ice, the authors have developed ice containing ozone micro-bubbles. In this paper, ice containing ozone micro-bubbles formed by adding surfactant was kept for a desired time at a desired constant temperature without melting of the ice. The concentration of ozone micro-bubbles fixed in ice and the ozone gas concentration released from ice by melting were measured to investigate the ozone decomposition rate due to fixation in ice and the characteristics of the released concentration. Furthermore, the influences of surfactant on both concentrations were examined.     

受阻胺类固化剂固化环氧树脂的固化过程研究

通过差示扫描量热法(DSC)研究了几种不同基团取代脂肪族二元胺中的一个氢原子所形成的三官能度脂肪胺类固化剂固化环氧树脂的过程。分别使用Kissinger模型和Flynn-Wall-Ozawa模型对三种固化剂固化环氧树脂的动力学参数进行了计算。结果表明,三种固化剂固化环氧树脂的反应活性大小比较为:N-乙基乙二胺≈N-甲基乙二胺>N-苯基乙二胺。这不仅与得到的固化度-时间曲线结果相符,并且可以从结构的角度以电子效应,共轭效应和位阻效应理论给予合理解释。N-苯基乙二胺中,由于苯环的刚性较大,空间位阻效应尤为明显,活化能(E a)在固化度(α)=0.5时为61.99 kJ/mol,而在α=0.9时达到97.63 kJ/mol,随着固化度的增加反应活性迅速下降,因此该类固化剂固化过程中更需要控制工艺。     

[数据建模]面向智能生产维护的大数据建模分析方法

从当前的车间离散制造过程大数据应用特性出发，基于车间智能代理，实现了制造过程大数据的采集、处理分析，建立的制造过程大数据采集模型可实现对工序数据集的封装。采用基于聚类的方法对工序数据集进行离群点检测标记，实现对制造资源生产状态的描述与量化。建立了工序级数据特征模型，结合改进的相似元分析方法，对工序过程数据多层次包装描述和制造工序的相似性进行了评价。开发了车间底层智能管理系统，并将其运用于企业生产实际。     

基于Agent的信任的情感劝说决策模型

基于智能体(Agent)的信任的情感劝说能更好地发挥Agent的各项人工智能优势。针对其中决策模型建立不够的现状,首先基于Agent的自身性格、情感衰减及外界刺激,运用大五性格、OCC情感模型及情感强度第一定律等,构建了相应的情感强度算法;其次考虑交易质量对信任的反馈,综合运用语言评价法、三角模糊数及将模糊数转换为具体分数(CFCS)算法,对Agent的交易质量进行模糊化和去模糊化处理,构建了相应的交易质量算法;再次对已有的信任更新因子进行改进后,构建了Agent的信任更新算法;最后在构建以上算法的基础上,采用交互式多准则决策(TODIM)法建立了基于Agent的信任的情感劝说决策模型,并采用熵权法对模型中的属性权重进行设定。为验证以上研究的合理性和有效性,设计了相应算例进行阐述,分析了不同参数取不同值得到的决策结果,并与相关研究进行了比较分析,得出了相应结论和管理启示,进一步归纳了不足之处和未来研究方向。     

Toward a new generic behavior model for human centered system simulation

The simulation of Human Centered Systems (HCS) has attracted increasingly attention in recent research. These systems involve individuals playing key roles, such as workers in manufacturing systems, soldiers in military operations, and investors in stock markets. The complexity simulating such systems is due to the need for modeling individual and group behavior and the integration of psychological and socio-technical aspects that can affect individual and HCS global performance. Although several models have been proposed to simulate such systems, most of them suffer from limitations pertaining to the integration of some factors, an inadequacy that will be discussed and elaborated on in this paper. The current study presents a new model for HCS simulation based on recent social and psychological theories. A model implementation example involving the simulation of a manufacturing system, considered as a HCS, is presented.     

T-ZnOw/TLCP抗静电复合材料的制备与性能研究

以热致液晶高分子(TLCP)为基体,四针状氧化锌晶须(T-ZnOw)为导电填料,用单螺杆挤出机制备抗静电性T-ZnOw/TLCP复合材料,分别探讨了偶联剂对T-ZnOw的影响;T-ZnOw用量对T-ZnOw/TLCP复合材料表面电阻率、热变形温度、力学性能的影响。结果表明,T-ZnOw经过偶联剂改性后,随着其用量的增加,复合材料表面电阻率显著下降,热变形温度、拉伸强度、冲击强度均呈现先增大后减小的趋势。     

改性氢氧化镁的制备及在HDPE中的应用

以低品位菱镁矿为原料,γ-(甲基丙烯酰氧)丙基三甲氧基硅烷(KH570)为改性剂,通过蒸氨-沉镁改性-水热过程制备改性氢氧化镁[KH570-Mg(OH)₂],改性剂在Mg(OH)₂生成体系中加入。采用吸油值、活化指数、红外光谱(FTIR)、X射线衍射(XRD)等手段对改性结果进行表征,在改性时间为60min,改性剂用量为2%,并进行水热处理工艺制备的Mg(OH)₂改性效果最好,改性后Mg(OH)₂活化指数达到了98.75%,吸油值为53.05 mL/100g,接触角为149°。整个工艺连续且可循环,为制备疏水性Mg(OH)₂工业生产提供实验基础。并将改性前后的Mg(OH)2应用于高密度聚乙烯(HDPE),测试其阻燃和力学性能,结果表明,KH570-Mg(OH)₂添加量为40%时,HDPE/KH570-Mg(OH)₂复合材料的氧指数为29.5%,达到阻燃级别,拉伸强度为18.40 MPa,断裂伸长率为217.11%,验证了改性后的Mg(OH)2与HDPE有较好的相容性。