Effect of K1, K2 Anti-Bacterial Agents on Tobacco Ralstonia Solanacearum

The tobacco Ralstonia Solanacearum were both cultured on nutrient agar plates and inoculated in seedling stage of tobacco, then treated with K1 and K2, two anti-bacterial agents, at a serial con-centrations to study their inhibitory efficiency. The result indicated that K1 can inhibit R. Solanacearum growth entirely, at the concentration range from 1/50 to 1/5000. K2 can reach the same result at the concentration range from 1/50 to 1/50000. Compared with the control plates, K1, at the concentration 1/50000, had no significant differences, and the average number of colony per plate was 112-115. The immature tobacco shown wilt as soon as inoculated with R. Solanacearum, and recovered gradually after using K1, K2. The densities of microbial suspension, handled by K1, K2 within 10 hs, were both significantly lower than the controlled ones. The optical microscopy also shown that handled microbial body differed from the controlled, whose body was regular short, rod shape as opposed to the handled ones with irregular rod shape and damaged body. All the results indicated that K1 and K2 both had inhibitory effects on tobacco R. Solanacearum, and K2 was more efficient than K1.

基于Agent人工智能的异构网络多重覆盖节点入侵检测系统设计

异构网络具有结构复杂、多重覆盖面积大等特征,使得网络入侵检测较为隐蔽,威胁网络运行的安全性;为此,对基于Agent人工智能的异构网络多重覆盖节点入侵检测系统进行了研究;通过检测Agent和通信Agent装设主机Agent,以Cisco Stealthwatch流量传感器作为异构网络传感器检测攻击行为,采用STM32L151RDT664位微控制器传输批量数据,由MAX3232芯片实现系统电平转化,实现硬件系统设计;软件部分设计入侵检测标准,采用传感器设备捕获网络实时数据,通过Agent技术解析异构网络协议并提取数据运行特征,综合考虑协议解析结果及与检测标准匹配度,实现异构网络多重覆盖节点入侵检测;经实验测试表明,基于Agent人工智能的异构网络多重覆盖节点入侵检测系统入侵行为的漏检率和入侵类型误检率的平均值仅为6%和5%,能够有效提高检测精度,减小检测误差。

流程型生产安全数据流的多Agent节点协同分流优化方法

为实现流程型生产安全监测系统的及时、准确决策,结合数据流采集与隐患识别过程,分析非关键数据的冗余、关键数据的缺失和数据计算时延较大的问题,提出基于多Agent的流程型生产安全数据流网络分流调度规划方法和节点流量划分识别机制。研究结果表明:相较于分簇传输方法,数据流网络分流调度方法可以实现关键隐患数据更高的传输成功率;相较于常规的复杂事件处理方法,本文提出的流量划分识别机制在2种类型数据集上实现隐患事件识别均有更低的计算时延。研究结果可为流程行业安全生产数字化管控模式和数据高质量获取提供参考。

基于AI agent的6G内生智能技术框架及其应用

未来6G网络将内生支持通信和AI一体化服务,赋能丰富多彩的新业务,支撑社会高效可持续发展。为此,借鉴了IT行业AI Agent的应用范式,基于电信应用场景创新地提出了6G AI Agent技术框架的三大设计理念,包括多模型融合、定制化Agent和插件式环境交互,并基于该理念构建了6G AI Agent技术框架。通过环境交互层、Agent引擎层、模型调度层、模型基座层交互协同,实现了自主环境感知、自主任务生成和自主执行任务的能力。此外,以移动网络的智能感知任务为例,探索了AI Agent的使用场景及价值,为AI新技术在电信领域发展提供了新的思路和技术支撑。

基于免疫Agent的电力电缆线路故障检测系统

为区别故障信号、非故障信号,实现对电缆线路故障的准确检测,设计了基于免疫Agent的电力电缆线路故障检测系统。利用前端检测电路为故障分析模块、测距方式切换模块提供电量传输信号,完成系统前端检测装置的连接。按照免疫Agent检测原理提取电力电缆线路故障信号的特征,联合已获取信号对象,求解检测插值指标的具体数值。结合各级硬件设备结构,完成系统设计。实验结果表明,该系统可同时检测波频为10~20 Hz、40~50 Hz、60~70 Hz的故障信号与波频为20~30 Hz、30~40 Hz、50~60 Hz的非故障信号,可以在精准辨别故障与非故障信号的同时,实现对电力电缆线路故障的准确检测。

基于多Agent的服装可持续消费行为建模与仿真

“双碳”目标下,随着服装消费引起的气候变化和环境污染等生态问题日益严重,迫切需要重新思考和重塑服装消费模式,推动服装消费向质量型、低碳化转变,从而实现服装产业的绿色转型。文章运用多Agent建模方法,构建服装可持续消费行为仿真模型,采用Netlogo平台进行仿真实验,探求服装可持续消费行为的影响机理。结果表明,消费者可持续消费认知、产品可持续性和设施供应条件均显著促进可持续消费行为;但不同强度下,产品可持续性和设施供应条件的推动作用存在差异。基于此,从可持续消费认知和供应条件两方面提出了引导中国消费者服装可持续消费行为的建议和对策。

基于多Agent深度强化学习的无人机协作规划方法

人机协作控制是多无人机任务规划的重要方式。考虑多无人机任务环境协同解释和策略控制一致性需求,提出基于多Agent深度强化学习的无人机协作规划方法。依据任务知识和行为状态,构建基于任务分配Agent的任务规划器,生成人机交互的相互依赖关系;设计一种深度学习强化方法,解决群体行为最优策略和协同控制方法,并利用混合主动行为选择机制评估学习策略。实验结果表明:作为人机交互实例,所提方法通过深度强化学习使群体全局联合动作表现较好,学习速度和稳定性均能优于确定性策略梯度方法。同时,在跟随、自主和混合主动3种模式比较下,可以较好地控制无人机飞行路径和任务,为无人机集群任务执行提供了智能决策依据。

一种基于Agent的任务自生成方法

任务自生成是基于预先设定的业务工作流完成任务执行流程制定的一个过程,任务执行流程制定主要包括任务执行步骤选取与任务执行步骤相应的解决方案选取。提出一种基于Agent的任务自生成方法来解决基于传统的业务工作流的流程制定机制,在面对有差异性需求的不同任务时,因其不能做出任务执行流程制定的自适应性改变而在处理任务时出现处理速度慢、结果不准确等问题。利用Agent的环境感知能力实现对任务需求信息和任务自生成系统环境的感知,利用Agent的自主行为决策能力结合感知的信息实现对任务执行流程制定工作的自主管理,在真实应用场景测试中,通过一系列比较实验验证了这种方法在处理速度和准确度方面具有明显优势。

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。

基于Multi-Agent的水电站变压器故障诊断系统

为了精准、快速完成水电站变压器的故障诊断,设计基于Multi-Agent的水电站变压器故障诊断系统。变压器状态监控agent将检测到的变压器故障信息发送给系统管理agent,系统管理agent通过通信agent将变压器故障信息发送给变压器故障诊断agent,变压器故障诊断agent利用小波变换方法提取变压器故障特征,并将其作为IFOA-SVM模型输入,完成变压器故障分类后,获取变压器故障诊断结果,该结果通过通信agent显示给用户。实验表明,该系统可有效诊断变压器故障诊断,诊断成功率受系统故障信息丢失率的影响较小,诊断耗时、耗能小,并具有较高故障诊断成功率。

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。

Patent-based technological developments and surfactants application of lithium-ion batteries fire-extinguishing agent

While newer,more efficient Lithium-ion batteries(LIBs)and extinguishing agents have been developed to reduce the occurrence of thermal runaway accidents,there is still a scarcity of research focused on the application of surfactants in different LIBs extinguishing agents,particularly in terms of patented technologies.The aim of this review paper is to provide an overview of the technological progress of LIBs and LIBs extinguishing agents in terms of patents in Korea,Japan,Europe,the United States,China,etc.The initial part of this review paper is sort out LIBs technology development in different regions.In addition,to compare LIBs extinguishing agent progress and challenges of liquid,solid,combination of multiple,and microencapsulated.The subsequent section of this review focuses on an in-depth analysis dedicated to the efficiency and challenges faced by the surfactants corresponding design principles of LIBs extinguishing agents,such as nonionic and anionic surfactants.A total of 451,760 LIBs-related patent and 20 LIBs-fire-extinguishing agent-related patent were included in the analyses.The extinguishing effect,cooling performance,and anti-recombustion on different agents have been highlighted.After a comprehensive comparison of these agents,this review suggests that temperature-sensitive hydrogel extinguishing agent is ideal for the effective control of LIBs fire.The progress and challenges of surfactants have been extensively examined,focusing on key factors such as surface activity,thermal stability,foaming properties,environmental friendliness,and electrical conductivity.Moreover,it is crucial to emphasize that the selection of a suitable surfactant must align with the extinguishing strategy of the extinguishing agent for optimal firefighting effectiveness.

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。

General anesthetic agents induce neurotoxicity through oligodendrocytes in the developing brain

General anesthetic agents can impact brain function through interactions with neurons and their effects on glial cells.Oligodendrocytes perform essential roles in the central nervous system,including myelin sheath formation,axonal metabolism,and neuroplasticity regulation.They are particularly vulnerable to the effects of general anesthetic agents resulting in impaired proliferation,differentiation,and apoptosis.Neurologists are increasingly interested in the effects of general anesthetic agents on oligodendrocytes.These agents not only act on the surface receptors of oligodendrocytes to elicit neuroinflammation through modulation of signaling pathways,but also disrupt metabolic processes and alter the expression of genes involved in oligodendrocyte development and function.In this review,we summarize the effects of general anesthetic agents on oligodendrocytes.We anticipate that future research will continue to explore these effects and develop strategies to decrease the incidence of adverse reactions associated with the use of general anesthetic agents.

意图与Agent承诺

理性的Agent对自己行为的控制依赖于自己的意图。意图具有两面性,它与意图行动和协调规划相联系。意图具有复杂的功能作用和相关规范特征,它们构成了意图的承诺特征。承诺有两个维度:意志的维度和以推理为中心的维度。承诺的意志维度的描述性说明包括:面向当前的意图在控制当下行为中的典型作用。承诺的规范性说明包括规范和“与这些作用相关的”理性规范。承诺既与“面向当前的意图在控制当前行为中的”作用有关,也与面向未来的意图“在意图的产生和意图的执行之间的过渡过程中的”作用有关。

General anesthetic agents induce neurotoxicity through astrocytes

Neuroscientists have recognized the importance of astrocytes in regulating neurological function and their influence on the release of glial transmitters.Few studies,however,have focused on the effects of general anesthetic agents on neuroglia or astrocytes.Astrocytes can also be an important target of general anesthetic agents as they exert not only sedative,analgesic,and amnesic effects but also mediate general anesthetic-induced neurotoxicity and postoperative cognitive dysfunction.Here,we analyzed recent advances in understanding the mechanism of general anesthetic agents on astrocytes,and found that exposure to general anesthetic agents will destroy the morphology and proliferation of astrocytes,in addition to acting on the receptors on their surface,which not only affect Ca^(2+)signaling,inhibit the release of brain-derived neurotrophic factor and lactate from astrocytes,but are even involved in the regulation of the pro-and anti-inflammatory processes of astrocytes.These would obviously affect the communication between astrocytes as well as between astrocytes and neighboring neurons,other neuroglia,and vascular cells.In this review,we summarize how general anesthetic agents act on neurons via astrocytes,and explore potential mechanisms of action of general anesthetic agents on the nervous system.We hope that this review will provide a new direction for mitigating the neurotoxicity of general anesthetic agents.

基于Agent的跨境电子商务课程智能教学系统

当前,跨境电子商务课程智能教学系统功能模块多为单向形式,可控教学覆盖范围较小,导致平均成绩优化比下降,为此提出对基于Agent的跨境电子商务课程智能教学系统的设计与分析。首先,构建通信接口,并接入驱动装置,完成系统硬件的设计;其次,采用多目标的方式,扩大可控教学覆盖范围;最后,设计多目标Agent跨境电子商务课程教学功能模块,关联跨境电子商务教学知识库,完成系统软件设计。测试结果表明:针对选定的6个班级进行测试对比,最终得出的平均成绩优化比均可以达到20%以上,说明在Agent的辅助下,文章设计的系统应用效果更佳,具有实际的应用价值。

基于多Agent的含DG智能配电网差动保护自愈方法

含DG智能配电网中的潮流方向随分布式发电的变化而变化,导致配电网故障自愈的效果难以保障,因此,提出基于多Agent的含DG智能配电网差动保护自愈方法。通过构建基于多Agent的MAS体系结构,将其应用于含DG智能配电网差动保护和自愈的控制,在MAS体系结构的数据处理层完成差动保护,并在MAS体系结构的协调控制层实现故障自愈。结果表明,该设计方法不仅可以恢复配电网故障区域的电力,而且在操作次数、线损及功率因数方面可达到较理想的效果。

基于AI Agent的高校图书推荐架构研究

随着人工智能技术的飞速发展,大语言模型和AI Agent在教育领域的应用逐渐深入。然而,高校的图书推荐系统局限于传统的以“书名”为核心的推荐算法,极大限制了优质图书资源的有效利用。研究旨在设计并实现一个基于AI Agent的图书推荐架构,为高校图书馆提供一个以知识点为颗粒度的图书信息检索和推荐方案,帮助提高图书馆书籍的检索准确性和推荐有效性。

Improvement Mechanism of Adhesion Performance of Anti-stripping Agents and Coupling Agents on Asphalt-Aggregate Interface Based on Molecular Dynamics

This study examined the mechanisms for improving the adhesion performance of the asphalt-aggregate interface with two anti-stripping agents and two coupling agents.The investigation of contact behavior between various asphalt-aggregate surfaces was conducted using molecular dynamics(MD)simulations.The interaction energy and the relative concentration distribution were employed as the parameters to analyze the enhancement mechanisms of anti-stripping agents and coupling agents on the asphalt-aggregate interface.Results indicated that the adhesion at the asphalt-aggregate interface could be strengthened by both anti-stripping agents and coupling agents.Anti-stripping agents primarily improve adhesion through the reinforcement of electrostatic attraction,while coupling agents primarily upgrade adhesion by strengthening the van der Waals.Hence,the molecular dynamics modeling and calculation techniques presented in this study can be utilized to elucidate the development mechanism of the asphalt-aggregate interface through the use of anti-stripping agents and coupling agents.