基于人工神经网络的振动压路机智能控制研究

在分析道路施工对振动压路机智能控制要求和现有控制方法的基础上,提出了振动压路机神经网络智能控制的方法,简介了该方法的系统结构与实现原理,重点研究了包括评价网络和动作网络在内的振动压路机神经网络控制模型与算法。仿真试验结果说明了采用神经网络模型进行振动压路机控制的可行性与正确性。

一种基于深度传感器的人体动作识别方法

本文设计了一种基于深度传感器的动作识别系统。该系统采用Kinect深度传感器对采集的人体动作视频进行特征提取,搭建了一个单层的BP神经网络进行训练和识别。实验表明,该系统在少样本训练的情况下,依然能够较准确地对站立、坐下、挥手与蹲下4个动作进行识别。该方法在数据维度上引入了第三维"深度信息",所以在网络训练时的速度和识别的响应速度要优于常规方法,也为人体动作识别提供了一种思路。

基于时空信息融合的时序动作定位

时序动作定位任务需要识别出一段长视频中的动作类别以及动作的起止时间,候选区域的选择是影响到识别效果和效率的重要因素。提出一种基于时空特征融合的候选区域提取网络,充分利用视频分割段的时间特征和空间特征来判断是否为候选区域。接着将候选区域输入到训练的CDC网络中进行帧级粒度上的动作分类。最后训练动作状态检测网络,对得到的候选区域进行修补,从而可以得到更为精确的动作发生的时间区域。在THUMOS'14数据集上进行实验,结果证明该方法可以有效地进行未剪辑视频的时序动作定位,相对现有方法达到了较高的精度。

关于公路主枢纽建设和枢纽网络系统运作机制的设想

根据交通部“三主一支持”的战略设想、公路主枢纽建设已进入实施阶段，形成以资本为纽带，以客货运两站车流为载体，以股份制为主要形式的集约化，集团公、网络化运行机制。

基于深度强化学习的蒸汽发生器水位控制

针对蒸汽发生器精确建模困难和低工况下控制性能差的问题,本文提出了一种基于深度强化学习优化的智能分层(IH)控制器。使用串级PI控制器作为初级控制器,用于直接控制水位。高级控制器采用经过深度强化学习优化的智能体控制器,负责实时对串级PI的参数进行优化,以便获得更好的控制性能。在高级控制器智能体的训练过程中,通过构建状态信息和奖励函数并采用深度残差神经网络逼近作为Q函数和策略函数的逼近器,获得了较好的泛化性能。结果表明:在不同的功率水平下,智能分层方法不仅对蒸汽发生器水位控制具有良好的跟踪能力,而且还具有很好的抗干扰能力。通过仿真验证了控制器的有效性。

基于目标导向行为和空间拓扑记忆的视觉导航方法

针对在具有动态因素且视觉丰富环境中的导航问题,受路标机制空间记忆方式启发,提出一种可同步学习目标导向行为和记忆空间结构的视觉导航方法.首先,为直接从原始输入中学习控制策略,以深度强化学习为基本导航框架,同时添加碰撞预测作为模型辅助任务;然后,在智能体学习导航过程中,利用时间相关性网络祛除冗余观测及寻找导航节点,实现通过情景记忆递增描述环境结构;最后,将空间拓扑地图作为路径规划模块集成到模型中,并结合动作网络用于获取更加通用的导航方法.实验在3D仿真环境DMlab中进行,实验结果表明,本文方法可从视觉输入中学习目标导向行为,在所有测试环境中均展现出更高效的学习方法和导航策略,同时减少构建地图所需数据量;而在包含动态堵塞的环境中,该模型可使用拓扑地图动态规划路径,从而引导绕路行为完成导航任务,展现出良好的环境适应性.

基于强化学习的高速飞行器巡航段高度控制

针对模型参数不确定情况下的高速飞行器高度控制问题,本文提出了一种基于强化学习的高度控制算法。在给出高速飞行器纵向运动学与动力学模型的基础上,将飞行器纵向模型分解成高度子系统和姿态子系统,并对高度子系统进行近似处理,使其具有与姿态子系统模型相同的形式。在此基础上,分别对高度子系统和姿态子系统进行基于强化学习的高度控制律和姿态控制律设计。最后分别对标称模型和气动参数极限拉偏情况下进行仿真实验。结果表明:所提出的方法对建模不确定具有很好的控制效果。

《神雕侠侣》日常任务等攻略

完美世界旗下首款2D回合动作网游——《神雕侠侣》将于9月20日12：00正式开启震撼公测。游戏基于先进的Valhalla引擎，从画面、特效、战斗打击感与爽快度以及首创自由轻功等众多玩法，为玩家带来从未有过的游戏体验。本次就为大家总结了游戏中能稳定高效获得经验的几种主要途径，尤其是刚进入《神雕侠侣》的诸位大侠们，一定会对你的大侠之旅有所帮助的。

5G Features from Operation Perspective and Fundamental Performance Validation by Field Trial

The fast deployment and penetration of 4G has cultivated human behaviors on mobile data consumption, leading to explosive growth in mobile traffic and stimulating new requirements on the capabilities of mobile networks. To meet the requirements of mobile networks toward year 2020, the next genera- tion of mobile networks （termed as IMT-2020, or 5G） is designed to support 100 Mbps-1 Gbps user-experienced data rate, 1 ms radio transmission latency, and 1 million connec- tions per square kilometer. Recalling the vision and requirements of 5G targeting for commer- cial launch in 2020, this article overviews the key features of 5G and compares with those of 4G, and reports the world first field trials conducted to validate the key performance of 5G radio interface in 3.SGHz band. The trial results show that a 1 ms transmission latency and 1 Gbps data rate are achievable.

焦虑状态前额叶皮层动作电位因果网络连接模式的研究

焦虑症是常见的情绪障碍类疾病,严重影响着患者正常生活并危害其身心健康。前额叶皮层是焦虑的重要责任脑区。应用清醒动物在体多通道微电极阵列记录技术可以同时获取大鼠在高架十字迷宫(EPM)焦虑状态的动作电位和行为学数据。基于极大似然估计方法(MLE)构建动作电位因果网络,计算网络连接强度和全局效率,定量表征大鼠焦虑状态内侧前额叶皮层动作电位因果网络连接模式。研究发现大鼠在高架十字迷宫封闭臂的进入次数(44.13±6.99)和停留时间(439.76±50.43)s,均高于开放臂的进入次数(16.50±3.25,P<0.001)和停留时间[(160.23±48.22)s,P<0.001]。封闭臂中的动作电位因果网络连接强度(0.017 3±0.003 6)和全局效率(0.044 2±0.012 8),均高于开放臂的网络连接强度(0.010 4±0.003 2,P<0.01)和全局效率(0.034 8±0.011 4,P<0.001),从封闭臂走向开放臂过程中连接强度与全局效率减弱。结果表明内侧前额叶皮层动作电位因果网络变化与焦虑状态有关,该结果可为研究焦虑状态的前额叶皮层脑网络机制提供支持。

Hardware Neural Networks Controlled MEMS Rotational Actuators and Application to Micro Robot

Hardware neural networks controlled rotational actuators and application to an insect type micro robot are reported in this paper. Millimeter size rotational actuators are fabricated by combining MEMS （Micro Electro Mechanical System） technology and shape memory alloy based artificial muscle wires. The actuator is composed of a pair of disk rotators and each rotor is suspended by four artificial muscle wires that are connected to the silicon frame. The rotational motion is generated by flowing the electrical current to each wire successively. Two actuators of different sizes are fabricated. The large actuator shows the displacement of 0.5 mm at the cycle time of 4 s. The small actuator shows 0.3 mm at 2 s. For controlling the actuator, the hardware neural networks are used. The hardware neural networks are composed of electrical circuits imitating cell bodies, excitatory synapses and inhibitory synapses. Four signal ports are extracted from four pairs of excitatory and inhibitory neurons and they are connected to the actuator. The small actuator is applied to the robot and built in the mid body of the robot. The shaft of the actuator is connected to the link mechanisms that transform the rotational motion to the locomotion. The appearance dimensions of the robot are 4.0, 2.7, 2.5 mm width, length and height. The robot performs forward and backward foot step like insects. The speed is 26.4 mm·min^-1 and the stepping width is 0.88 mm. Also, the robot changes the direction by external trigger pulses.

A Novel Smart Forwarding Scheme in LTE-Advanced Networks

Long Term Evolution(LTE) and IEEE 802.16 WiMAX are competing access network technologies adopted in 4G wireless networks in recent years.LTE complies with3 GPP standards whereas 802.16 WiMAX is regulated by the Institute of Electrical and Electronics Engineers(IEEE).Although WiMAX is already operating commercially in Taiwan,the system is limited to an independent new system that is incompatible with the current 3G system.Hence,the cost of implementing the WiMAX system is relatively high,this being an impediment to its rapid uptake and widespread use.On the other hand,LTE conforms to 3GPP that is supported by telecommunication manufacturers and operators and is,moreover,backward compatible with 3G/UMTS cellular systems.The LTE specifications define how user equipment(UE) connects and communicates with evolved Node B(eNB) base stations.The enhanced version,LTE-Advanced,adds a new entity called the relay node(RN) to widen service coverage,although this change has resulted in a more complex architecture.Mobility management and data forwarding are essential components in wireless mobile networking.This paper focuses on the efficient handover procedure in LTE-Advanced networks,and proposes a Smart Forwarding mechanism to improve the handover performance.Simulation studies show that the proposed Smart Forwarding scheme employs a better operational transmission path that effectively reduces handover latency and signal overhead.

Social-Aware Cooperation Among Mobile Terminals for Wireless Downlink Transmission

This paper investigates the social-aware cooperation(SAC) among mobile terminals(MTs), motivated by the fact that modern smart devices have much improved context awareness. Aware of the social ties, the cooperative network contains two layers of property: social and physical. In order to observe how the social awareness benefit the cooperation performance, we first formulate the social ties between MTs into parameters that can describe the cooperative behaviors by taking the mobility feature into account, defined as the conviction-approval-suspicion(CAS) model. Limited by the processing capability, partner selection is of great practical significance. To this end, the social-aware partner selection strategy is analyzed, and a significant superiority is observed compared to social-unaware selection. By analyzing the cooperative throughput, an explicit relationship between the degrees-of-freedom gain and the social-physical property is finally derived. Simulation results validate the theoretical analysis.

Dynamic Evolution with Limited Learning Information on a Small-World Network

This paper investigates the dynamic evolution with limited learning information on a small-world network.In the system, the information among the interaction players is not very lucid, and the players are not allowed to inspectthe profit collected by its neighbors, thus the focal player cannot choose randomly a neighbor or the wealthiest one andcompare its payoff to copy its strategy.It is assumed that the information acquainted by the player declines in theform of the exponential with the geographical distance between the players, and a parameter V is introduced to denotethe inspect-ability about the players.It is found that under the hospitable conditions, cooperation increases with therandomness and is inhibited by the large connectivity for the prisoner's dilemma; however, cooperation is maximal atthe moderate rewiring probability and is chaos with the connectivity for the snowdrift game.For the two games, theacuminous sight is in favor of the cooperation under the hospitable conditions; whereas, the myopic eyes are advantageousto cooperation and cooperation increases with the randomness under the hostile condition.

uSensing：a Universal Platform to Support Smartphones Accessing Wireless Sensor Networks

This paper presents a universal platform "uSensing" to support smartphones to communicate with sensor nodes in Wireless Sensor Networks (WSNs).Since phones have different CPU processers and operating systems,it is a challenge to merge these heterogeneities and develop such a universal platform.In this paper,we design both hardware and software to support the "universal" feature of uSensing:1) "uSD" card:an IEEE 802.15.4 physical communication card with SD interface;2) "uSinkWare":a WSNs middleware running on smartphones.Integrated with uSD card and uSinkWare,phones become mobile data sinks to access into WSNs and parse messages from sensor nodes.We demonstrate the proposed uSensing platform in a commercial smartphone to connect with our WSNs testbed,and validate that the smartphone has the same WSNs functions as commercial fixed sink.Additionally,we evaluate the performance of uSensing platform through measuring phone's CPU load and power consumption,and analyze the performance of these metrics theoretically.The results suggest that the phone-based mobile sink has enough capability to serve as a mobile sink of WSNs and can work up to twenty hours due to low power consumption.

You Blog： An Exploratory Study of EFL Students＇ Blogging Experience in the Thai-Japanese Weblog Exchange Project

Weblogs, writing logs on the web, have been popularly used as an instructional tool to facilitate, reading and writing instructions in ESL （English as a Second Language）/EFL （English as a Foreign Language） classrooms. This study employed weblogs to support the writing process of Thai EFL students and provided an opportunity for them to participate in a Thai-Japanese weblog exchange project. The results showed that student bloggers： （1） enjoyed blogging due to their perceived reading and writing incentives, the exchange of cultural knowledge, and networking; and （2） employed the writing process to complete their written assignments. Findings suggest that weblogs can foster EFL learners＇ motivation, development of reading and writing skills, and development of cultural knowledge, and facilitate their writing process via cyberspace.

分层决策多机空战对抗方法

在空战研究领域,战术决策旨在提高博弈对抗收益,进而提升战机攻击效率.现有战术决策算法大多基于规则方法设计,当应用于多机空战的复杂环境时则存在设计难度大,难以求解最优解等问题.本文提出一种分层决策多机空战对抗方法,首先,在训练初始阶段借鉴已有人类专家经验,指导模型训练;其次,根据战术动作类型设计分层动作决策网络,降低动作决策空间维度;最后,将训练产生的对抗经验按阶段分解,降低策略学习难度.在多机空战仿真环境中进行了实验验证,相比于现有多机空战决策方法,本文提出的方法在训练收敛性和决策性能方面均具有更好的表现.

A forward-inverse dynamics modeling framework for human musculoskeletal multibody system

Multibody musculoskeletal modeling of human gait has been proved helpful in investigating the pathology of musculoskeletal disorders.However,conventional inverse dynamics methods rely on external force sensors and cannot capture the nonlinear muscle behaviors.Meanwhile,the forward dynamics approach is computationally demanding and only suited for relatively simple tasks.This study proposed an integrated simulation methodology to fulfill the requirements of estimating foot-ground reaction force,tendon elasticity,and muscle recruitment optimization.A hybrid motion capture system,which combines the marker-based infrared device and markerless tracking through deep convolutional neural networks,was developed to track lower limb movements.The foot-ground reaction forces were determined by a contact model for soft materials,and its parameters were estimated using a two-step optimization method.The muscle recruitment problem was first resolved via a static optimization algorithm,and the obtained muscle activations were used as initial values for further simulation.A torque tracking procedure was then performed by minimizing the errors of joint torques calculated by musculotendon equilibrium equations and inverse dynamics.The proposed approach was validated against the electromyography measurements of a healthy subject during gait.The simulation framework provides a robust way of predicting joint torques,musculotendon forces,and muscle activations,which can be beneficial for understanding the biomechanics of normal and pathological gait.

ANALYSIS FOR GENE NETWORKS BASED ON LOGIC RELATIONSHIPS

The reverse construction and analysis of the networks of molecular interactions are essential for understanding their functions within cells. In this paper, a logic network model is constructed to investigate the complicated regulation mechanism of shoot genes of Arabidopsis Thaliana in response to stimuli. The dynamics of the complicated logic network is analyzed, discussed, and simulated. The simulation results show that the logic network of the active genes of shoot eventually evolves into eleven attractors under the stimuli, including five 1-periodic and six 2-periodic attractors. Our work provides valuable reference and guidance for biologists to understand and explain Arabidopsis＇ response to external stimuli by experiments.