摘要：A screw drive in-pipe robot is promising inspection equipment for small pipes. However, most of the existing screw drive in-pipe robots have problems of motion interference and slipping inside curved or irregular pipes. These problems result from the coupled relations among the steering motion, the motion speed and the load ability of the robot. In order to deal with the problems, the axiomatic design(AD) theory is applied to evaluate and analyze the existing designs. Then an uncoupled concept design based on the AD theory is proposed and the complete AD decomposition process is presented. After that, the proposed robot based on a tri-axial differential angle modulation mechanism is designed to realize the uncoupled concept. Finally, the uncoupled property is verified in a dynamics simulation system. The simulation results indicate that the motion speed, load ability and steering motion of the proposed robot can be adjusted individually compared with the robots that have inclining-angle-fixed rollers. Owing to the uncoupled design, the proposed robot can mechanically adapt to straight pipes and curved pipes with less roller slipping.

摘要：文章对公共空间中行为博弈的多重机制进行解析,揭示了人群行为作为空间形态要素的本质,分析了人群所构成的社会空间形态对个体行为决策的影响原理。结合课程教学中运用数字化感知工具开展的行为研究案例,阐述了基于空间占位和感官刺激的两种行为博弈模式,并将这两种模式拓展到人与物体、人与城市动物、人与智能机器三个不同的范畴,拓展了基于“空间-行为”的设计认知。

摘要：针对当前北斗卫星导航系统(BDS)精密卫星钟差预测模型精度不高、预测误差较大等问题,提出一种将鲸鱼优化算法(WOA)和径向基函数神经网络(RBF)相结合的钟差预测模型—鲸鱼算法优化的RBF组合模型(WOA-RBF):利用四分位法和分段线性插值法完成数据预处理,通过鲸鱼优化算法对RBF中的扩展速度和输出层线性权重进行寻优,得到最优参数,最终得到优化后的输出结果。实验结果表明:与二次多项式(QP)模型、灰色模型(GM)和径向基函数神经网络(RBF)模型相比,WOA-RBF模型优势明显。在预报时长为6 h时,均方根误差在0.25 ns以内;在预报时长为12 h时,均方根误差在0.27 ns以内,证明了WOA-RBF模型在精密卫星钟差短期预报中的准确性和可行性。

摘要：According to the theory of DBR, with the P-type DBR as an example, the electrical characteristics and optical reflection of the DBR are analyzed by studying the energy band structure with various graded region widths and doping densities. The width and doping density of graded region are decided through a comparative study. The P-type DBR of 980 nm VCSELs is designed with Al0.9Ga0.1As and Al0.1Ga0.9As selected as the high and low refractive index material for the DBR. The 980 nm bottom VCSELs, which consists of 30 pairs P-type DBR and 28 pairs N-type DBR, are then fabricated. In P-type DBR, the width of graded region is 0.02 μm and the uniformity doping concentration is 2.5×10^18cm^-3. Its reflectivity is 99.9%. In N-type DBR, the width of graded region is also 0.02 μm and the uniformity doping concen- tration is 2×10^18cm^-3. Its reflectivity is 99.3%. The I-V curve shows that the series resistance of the device is about 0.05Ω. According to the theory of DBR, with the P-type DBR as an example, the electrical characteristics and optical reflection of the DBR are analyzed by studying the energy band structure with various graded region widths and doping densities. The width and doping density of graded region are decided through a comparative study. The P-type DBR of 980 nm VCSELs is designed, with Al0.9Ga0.1As and Al0.1Ga0.9As selected as the high and low refractive index material for the DBR. The 980 nm bottom VCSELs, which consist of 30 pairs P-type DBR and 28 pairs N-type DBR, are then fabricated. In P-type DBR, the width of graded region is 0.02μm and the uniformity doping concentration is 2.5×10^18cm^-3. Its reflectivity is 99.9%. In N-type DBR, the width of graded region is also 0.02 μm and the uniformity doping concentration is 2×10^18cm^-3. Its refiectivity is 99.3%. The I-V curve shows that the series resistance of the device is about 0.05Ω.

摘要：对AI绘画的起源、发展及其在艺术设计领域的应用进行综述,探讨AI绘画与设计伦理之间的关系,并进行设计伦理理论指导下的AI绘画创新应用。

摘要：强暴雨常伴随雷电天气会导致输电线路、变电站和换流站绝缘子表面干、湿部位电场分布不均,易引发电场畸变,同时雷暴引起的雨柱会快速短接绝缘子伞间距,显著缩短电弧发展路径,致使线路跳闸并引起停电事故。为研究强暴雨条件下线路绝缘子雷电冲击放电特性和电弧发展过程,针对3种110 kV线路常用硅橡胶复合绝缘子开展了雷电冲击放电特性测试。结果表明:复合绝缘子雷电冲击电弧发展过程主要分为起弧、局部电弧发展、完全闪络,其放电发展时间及其持续时间均随降雨强度、雨水电导率的增大而降低;电弧平均发展速度随雨水电导率的增加呈幂指数关系上升,最大上升率高达43.8%,并随降雨强度增加呈指数式增大,且与绝缘子伞裙结构参数有关。复合绝缘子雷电冲击闪络电压随降雨强度增加呈负幂指数关系降低,伞间距较大的绝缘子闪络电压受降雨强度的影响愈加明显;雨闪电压随雨水电导率的增加而呈负幂指数下降。基于此,提出了降雨强度、雨水电导率综合影响的复合绝缘子雷电冲击闪络电压计算式,计算结果与试验结果一致性良好,偏差小于7%。研究结果为雷电活动频繁的强暴雨地区输电线路外绝缘设计、选择及防护提供了参考。

摘要：为了提高低温工作环境下808 nm半导体激光器的输出特性,深入研究了电光转换效率的温度特性。结合载流子泄漏抑制和器件串联电阻的优化考虑,从理论上深入分析了有源区量子阱内的载流子限制现象,提出针对低温工作环境下的势垒高度及相应的量子阱结构设计方法,包括势垒层的材料组分、厚度等重要参数的优化,极大地改善了器件在低温工作环境下的性能。采用优化后的外延结构,制备了腔长2 mm的半导体激光巴条。在工作温度-50℃、注入电流为600 A时,巴条输出功率达到799 W,电光转换效率为71%,斜率效率为1.34 W/A;注入电流为400 A时,器件达到最高电光转换效率73.5%,此时的载流子限制效率约为99%,串联电阻为0.43 mΩ;在-60~60℃温度范围内,中心波长随温度的漂移系数为0.248 nm/℃。

摘要：文本语义匹配是很多自然语言处理任务的基础.在很多场景中都需要文本语义匹配技术,如搜索、问答系统等.在实际运用场景中,对文本语义匹配的效率有很高的要求.虽然表征学习型语义匹配模型相较于交互型模型的准确率有所下降,但效率极高.而表征学习型语义匹配模型提升性能的关键是抽取具有高层语义特征的句向量.针对该问题,本文在ERINE模型的基础上,设计了特征融合模块及特征抽取模块,以获取具有多维语义特征的句向量,并通过设计语义预测的损失函数,进一步提升模型获取语义信息的性能,从而提高文本语义匹配的准确率.最终在百度千言文本相似度数据集上的准确率达到85.1%,表现出较好的性能.

摘要：鉴于施工现场中工人与施工机械及施工用具之间不合规的交互关系是引发安全事故的重要原因,提出了一种基于施工场景的视觉关系检测方法。首先,采用卷积神经网络搭建实体检测和关系检测分支,以提取出施工场景中的实体特征和关系特征;其次,构建视觉模块、语义模块和空间模块对提取出的特征进行学习,使网络充分感知和理解视觉信息、语义信息与空间信息;最后,设计了一种图形对比损失函数,以提高模型的视觉关系检测性能。在自制的施工场景关系检测数据集上的实验结果表明,本文方法实现了75.89%、77.64%、78.93%的R@20、R@50、R@100召回率,具有良好的视觉关系检测性能,能精准地检测出施工场景中的目标及其交互关系。

摘要：目前,养猪场母猪自动化喂料系统多采用发达国家母猪饲喂站的方式,虽然实现了自动喂料,但无法实现每头母猪按规定时间进食,且易造成交叉感染。为此,设计了一种采用电脑和可编程逻辑控制器PLC进行中央控制,通过控制喂料器运转时长将饲料依次加入各母猪单体栏内食槽,实现针对每头母猪的定时、定量精准喂饲的自动化喂料系统,并通过正交试验和方差分析确定了喂料器输料管直径、螺旋轴转速、物料输送仰角等各项参数。研究结果表明:饲喂系统可使投料误差控制在4%以内,实现了精准自动饲喂,确保了养猪场的正常生产和管理水平的提高。