基于Pyramid-Attention-U-Net深度学习模型的实时拓扑优化设计

拓扑优化设计可为现代工程提供具有卓越热学、力学及声学等多学科性能的创新型结构,移动变形组件法(Moving Morphable Components,MMC)以其独特的显示拓扑优化方法备受青睐,MMC法通过一系列可移动变形的显示组件间的移动、变形、重叠实现边界演化以完成结构优化的目的。该文利用椭圆形初始组件替代原有直线型骨架厚度二次变化组件进行拓扑优化。在减少设计变量次数的同时,可缩短一定的计算时间,但在实际计算中,随着初始组件单元数增加,中间迭代计算过程时间依旧相对较多,为精确实时获取拓扑优化构型,该文引入Pyramid-Attention-U-Net(PA-U-Net)深度学习模型加速优化设计,避免中间迭代计算过程。研究结果表明:该方法不仅在可忽略的计算时间内准确实时获取不同参数下的初始组件拓扑构型,而且准确率可达90.89%,高于其他深度学习网络模型。同时这种将深度学习与拓扑优化方法有机结合的形式在大型工程结构优化设计中具有广阔的应用前景。

基于FACAM-Efficient Net的通信信号调制方式识别

针对在复杂电磁空间中,通信信号识别方法存在噪声干扰、识别率不足、网络模型复杂等问题,提出基于高阶累积量可视化时频图像的FACAM-Efficient Net信号调制方式识别方法,通过高阶累积量时频分析方法,将不同调制的通信信号转换成可视化时频图像,用作网络模型的训练和测试,根据可视化时频图像的图像特征,提出FACAM频域通道混合注意力机制,并引入FACAM改进Efficient Net的网络主体结构,更好地提取时频图像的频域和通道的特征。实验结果显示,本文提出的调制识别方法在低信噪比下有较好的识别效果。

基于Time Petri Nets的实时系统资源冲突检测

Time Petri Nets在实时系统的建模和性能分析中得到广泛应用,而冲突是Petri网及其扩展模型的重要行为,解决冲突是正确分析模型动态行为的关键.目前随机Petri网、混合Petri网和区间速率连续Petri网的冲突检测方法由于没有考虑到时间约束因此无法在TPN网中使用.时间约束的引入使得Time Petri Nets模型的使能和触发语义比Petri网模型的语义复杂,冲突检测变得更加困难.为了计算冲突发生的时间和概率,首先根据时间约束,给出了变迁持续使能时延迟区间的计算方法,并证明了该方法的合理性和完备性;然后在此基础上定义并证明了Time Petri Nets模型中不冲突的检测方法;并提出了Time Petri Nets模型的冲突检测方法,给出了冲突时间区间和变迁实施概率的计算方法;最后通过实例验证说明了该方法的正确性和有效性.

基于Time Petri Net的实时系统冲撞检测与消解

time Petri net(TPN)在实时控制系统的建模中得到广泛应用,而冲撞是Petri网及其扩展模型的重要行为,解决冲撞是正确分析模型动态行为的关键.由于引入时间约束,使得TPN模型的使能和触发语义比Petri网模型的语义复杂,冲撞的检测及消解变得更加困难.首先根据时间约束,给出了变迁持续使能时延迟区间的计算方法,并证明了该方法的正确性;然后在此基础上定义并证明了TPN模型中冲撞的检测方法;给出了冲撞时间区间及修改时间约束的冲撞消解方法;最后通过实例验证说明了该方法的有效性和正确性.

基于Time Petri Nets的UML时序图分析

引入一种称为Clock的变迁,用来改进TPNs,讨论了如何将时序图转换为基于Clock变迁的TPNs,使这种TPNs能正确反映时序图的流程和时间约束。最后,利用普通Petri网的可达性分析技术对时序图模型进行了分析和验证。

A Simulation Engine for Stochastic Timed Petri Nets and Application to Emergency Healthcare Systems

In many service delivery systems,the quantity of available resources is often a decisive factor of service quality.Resources can be personnel,offices,devices,supplies,and so on,depending on the nature of the services a system provides.Although service computing has been an active research topic for decades,general approaches that assess the impact of resource provisioning on service quality matrices in a rigorous way remain to be seen.Petri nets have been a popular formalism for modeling systems exhibiting behaviors of competition and concurrency for almost a half century.Stochastic timed Petri nets(STPN),an extension to regular Petri nets,are a powerful tool for system performance evaluation.However,we did not find any single existing STPN software tool that supports all timed transition firing policies and server types,not to mention resource provisioning and requirement analysis.This paper presents a generic and resource oriented STPN simulation engine that provides all critical features necessary for the analysis of service delivery system quality vs.resource provisioning.The power of the simulation system is illustrated by an application to emergency health care systems.

The Relation between the Stabilization Problem for Discrete Event Systems Modeled with Timed Petri Nets via Lyapunov Methods and Max-Plus Algebra

A discrete event system is a dynamical system whose state evolves in time by the occurrence of events at possibly irregular time intervals. Timed Petri nets are a graphical and mathematical modeling tool applicable to discrete event systems in order to represent its states evolution where the timing at which the state changes is taken into consideration. One of the most important performance issues to be considered in a discrete event system is its stability. Lyapunov theory provides the required tools needed to aboard the stability and stabilization problems for discrete event systems modeled with timed Petri nets whose mathematical model is given in terms of difference equations. By proving stability one guarantees a bound on the discrete event systems state dynamics. When the system is unstable, a sufficient condition to stabilize the system is given. It is shown that it is possible to restrict the discrete event systems state space in such a way that boundedness is achieved. However, the restriction is not numerically precisely known. This inconvenience is overcome by considering a specific recurrence equation, in the max-plus algebra, which is assigned to the timed Petri net graphical model.

Timed Petri Net Models of Shared-Memory Bus-Based Multiprocessors

In shared-memory bus-based multiprocessors, when the number of processors grows, the processors spend an increasing amount of time waiting for access to the bus (and shared memory). This contention reduces the performance of processors and imposes a limitation of the number of processors that can be used efficiently in bus-based systems. Since the multi-processor’s performance depends upon many parameters which affect the performance in different ways, timed Petri nets are used to model shared-memory bus-based multiprocessors at the instruction execution level, and the developed models are used to study how the performance of processors changes with the number of processors in the system. The results illustrate very well the restriction on the number of processors imposed by the shared bus. All performance characteristics presented in this paper are obtained by discrete-event simulation of Petri net models.

An Extension of Contract Net Protocol With Real Time Constraints

The contract net protocol has developed to specify problem solving communication for nodes in a distributed problem solving. Task distribution is affected by a negotiation process,a discussion carried on between nodes with tasks to he executed and nodes that may be able to execute those tasks. In contract net protocol,once negotiation successes,tbe task execution is assumed to success. However,in real world,even though a task is awarded to successfully bidding nodes,it may be delayed. Such delay may badly propagate in whole system. Here,we introduce real-time constraints into contract net protocol to manage task execution for avoiding the, task's delay,or even though being delayed,the railure cannot propagate to whole system. In this paper,we first present a real-time contract net protocol which is an extension of contract net protocol with real-time constraints for distributed computing. Our proposition extends the basic negotiation protocol to negotiation and controlling execution or task. The controlling process is based on task deadline time,we also present an extension of the internode language of contract net protocol specification with real-time constraints.

多模块U-Net-BiLSTM网络驱动的滚动轴承寿命预测方法研究

针对滚动轴承寿命预测方法难以准确识别故障始发时刻(FPT)和提取时间序列深层特征的问题,提出了一种联合高精度FPT点和多模块U-Net-BiLSTM网络的滚动轴承寿命预测方法:对小波降噪后原始信号功率谱中每一时刻内所有频率成分进行累加求和,联合欧氏距离准则与3σ原则识别高精度FPT点;分别将残差块、池化层和归一化层引入编码器和解码器中实现多尺度特征融合,从而改进传统U-Net网络,有效提升了模型对时序信号的处理能力和预测速度。实验结果表明:相较于现有3种深度学习方法,具有更高的预测精度和更快的预测速度。

USING COLORED-TIMED PETRI NET FOR THE SOLUTION OF DEADLOCK IN A FLEXIBLE ASSEMBLY SYSTEM

A colored timed Petri net(CTPN)modeling technique is developed for flexible assembly system(FAS) In particular,one section is devoted to the deadlock nodel based on the Petri net The rule of the deadlock error recovery is given Through the analysis of the behavioral properties of the FAS with the deadlock recovery model,the deadlock problem would be solved in the system.

Assessment of the methods for determining net radiation at different time-scales of meteorological variables

When modeling the soil/atmosphere interaction,it is of paramount importance to determine the net radiation flux.There are two common calculation methods for this purpose.Method 1 relies on use of air temperature,while Method 2 relies on use of both air and soil temperatures.Nowadays,there has been no consensus on the application of these two methods.In this study,the half-hourly data of solar radiation recorded at an experimental embankment are used to calculate the net radiation and long-wave radiation at different time-scales(half-hourly,hourly,and daily) using the two methods.The results show that,compared with Method 2 which has been widely adopted in agronomical,geotechnical and geo-environmental applications.Method 1 is more feasible for its simplicity and accuracy at shorter time-scale.Moreover,in case of longer time-scale,daily for instance,less variations of net radiation and long-wave radiation are obtained,suggesting that no detailed soil temperature variations can be obtained.In other words,shorter time-scales are preferred in determining net radiation flux.

FINE-GRAINED DISTRIBUTED MULTIMEDIA SYNCHRONIZA-TION MODEL——ENHANCED FUZZY-TIMING PETRI NET

A fine grained distributed multimedia synchronization model——Enhanced Fuzzy timing Petri Net was proposed which is good at modeling indeterminacy and fuzzy. To satisfy the need of maximum tolerable jitter, the sufficient conditions are given in intra object synchronization. Method to find a proper granularity in inter object synchronization is also given to satisfy skew. Exceptions are detected and corrected as early as possible using restricted blocking method.

Hierarchical Colored Timed Petri Nets for Maintenance Process Modeling of Civil Aircraft

Civil aircraft maintenance process simulation model is an effective method for analyzing the maintainability of a civil aircraft. First, we present the Hierarchical Colored Timed Petri Nets for maintenance process modeling of civil aircraft. Then, we expound a general method of civil aircraft maintenance activities, determine the maintenance level for decomposition, and propose the methods of describing logic of relations between the maintenance activities based on Petri Net. Finally, a time Colored Petri multi-level network modeling and simulation procedures and steps are given with the maintenance example of the landing gear burst tire of a certain type of aircraft. The feasibility of the method is proved by the example.

Modeling of Fuel Elements Cycling System in Pebble Bed Reactor Based on Timed Places Control Petri Nets

Pebble bed reactors use cycling scheme of spherical fuel elements relying on fuel elements cycling system (FECS). The structure and control logic of FECS are very complex. Each control link has strict requirements on time and sequence. This increases the difficulties of description and analysis. In this paper, timed places control Petri nets (TPCPN) is applied for the modeling of FECS. On this basis the simulation of two important processes, namely uploading fuel elements into the core for the first time and emptying the core is finished by simulation software Arena. The results show that as TPCPN is able to describe different kinds of logic relationship and has time properties and control properties, it’s very suitable for the modeling and analysis of FECS.

基于STPA和FTPN的海上自主水面船舶航行实时风险评估

为实时监测海上自主水面船舶(MASS)航行过程风险,基于系统理论事故模型与过程(STAMP)建立MASS的安全控制结构,采用系统理论过程分析法(STPA)确定损失/事故和系统级危险,识别不安全控制行为并分析损失场景,构建系统状态转化过程模型;采用模糊时间Petri网(FTPN)建模,以设定的MASS航行场景得到相关模糊时间函数并推算FTPN的情态演进;引入新的风险水平表达式并通过系统实时损失/事故二维路径图来可视化系统的实时风险水平和系统不安全状态的转化路径。结果表明:设定的航行场景在当前时刻下,缺少安全水深输入、未更新避碰路径、航向航速不安全、搁浅是风险最高的系统不安全状态,并对应4条风险最高的转化路径;STPA驱动下的FTPN过程模型能全面评估MASS航行的实时风险水平,以系统实时损失/事故二维路径图作为可视化界面,用于监管MASS航行中不安全系统状态并描述其转化路径。

面向流程制造的数字孪生车间可视化监控系统研究

为了解决流程制造过程监控实时性差、准确性不足的问题,在数字孪生技术迅速发展且与制造业结合愈发紧密的背景下,提出基于生产要素数据、生产工艺流程与多工序设备实体对象的数字孪生车间可视化监控系统架构。从要素、状态、运行逻辑等多个维度对多工序耦合生产线实体对象进行孪生建模,基于OPC UA统一架构的数据采集传输,虚拟车间的实时驱动等关键技术,实现了流程制造设备状态、生产要素工艺全流程三维可视化监控。基于某流程制造企业制丝生产试验线开展了应用验证,为提升流程制造过程中全生产要素的预见性、联动性、自治性提供了方法和实现途径。

基于循环生成对抗网络的逆时偏移成像结果优化

在常规逆时偏移方法基础上,通过引入循环生成对抗网络(CycleGAN)发展一种基于循环生成对抗网络的逆时偏移成像结果优化方法。首先构建包含两个生成器和两个判别器的CycleGAN。在对抗损失、循环一致性损失函数的基础上,添加身份损失函数,以避免训练过度;然后,组建样本集来训练网络,使其学习常规逆时偏移成像结果和最小二乘逆时偏移成像结果之间的映射关系;最后,利用其他合成数据和实际资料测试网络效果。结果表明,提出的基于循环生成对抗网络的逆时偏移成像结果优化方法在获得高精度、高信噪比成像结果的同时有效地提高了计算效率。

2001-2020年内蒙古净生态系统生产力格局多时间尺度分析

掌握净生态系统生产力(NEP)时空格局对提高干旱/半干旱地区生态系统功能有重要意义。已有的NEP时空格局研究大多以年尺度开展分析,而NEP在多时间尺度上的特征差异尚不明晰。基于多源遥感、气象和地面实测数据,采用CASA模型、土壤呼吸地质统计模型(GSMSR)和土壤呼吸-土壤异养呼吸(Rs-Rh)关系模型耦合模拟内蒙古2001-2020年NEP,分析其年、季、月多时间尺度时空特征,并探讨8种不同植被NEP的多时间尺度特征差异。结果表明,1)内蒙古年尺度NEP的空间分布格局稳定,从东北向西南递减,这一格局与春夏秋3季及植被生长期的3-10月一致,而冬季植被进入休眠期使得空间差异显著减小。2)内蒙古多时间尺度NEP年际变化趋势有所不同:年尺度上,内蒙古总NEP呈波动上升趋势,年际变化率为C 3.75 Tg·a^(-1);季尺度上,夏季增长趋势最大,占全年增长的41.6%,春秋两季对NEP的增长也起到至关重要的作用,分别占比34.9%和23.3%,冬季对NEP增长贡献非常有限;月尺度上,NEP年内变化与植被生长物候周期较为接近,1月和12月年际NEP为减少趋势,其余月份年际NEP均为上升趋势,其中9月增长趋势最大,占全年增长的19.3%。3)不同植被类型NEP年际趋势存在差异,5种植被类型年NEP呈增长趋势,3种呈下降趋势;草地在季尺度年际变化中均保持增长,在夏季最高,而灌木林在季尺度年际变化中均为降低趋势,夏季降幅最大;月尺度年际变化中灌木林均为降低趋势,1月降幅最大。该研究能够为明晰区域碳循环及改善生态系统功能提供科学依据。

Dueling Double DQN在基于MPN混流制造车间实时调度中的应用

针对传统算法难以适应当今大规模,多资源约束的混流制造实时调度场景的问题,提出一个可缩短最终完工时间的双层决斗DQN(D3QN)算法实时调度框架。在该框架内,制造车间经压缩建模成制造Petri网(MPN)模型,通过新的奖励机制反复推演MPN仿真生产过程,评估排产收益并产生大量样本数据,将数据中多维生产特征信息矩阵作为车间状态输入多通道卷积神经网络,采用D3QN算法训练网络模型,一旦网络模型收敛至最优价值函数,即可调用该网络模型结合在线匹配执行机制,快速匹配车间生产状态,执行最优工件排产变迁的决策动作。实验数据表明:在最佳超参数设置下,使用D3QN算法训练的网络模型,其求解性能和响应速度满足混流制造车间实时调度需求。