摘要：With the prompt development in intellectualization nowadays, the smart materials with multifunctionality or multi-responsiveness are highly expected. But it is a big challenge to integrate the different actuating units into a single system in a synergy pattern. Herein, we put forward a new strategy to develop the polyurethane networks which can present shape-memory effect and self-healing effect in independent way as well as simultaneous acting mode. To realize this goal, poly(tetremethylene ether) glycol was chosen as the soft segment to ensure the polymer chains a good mobility, and disulfide bond as the dynamic covalent bond was embedded in the backbone of polyurethane to endow it with desirable self-healing capacity under mild condition. Moreover, a rational control of the architecture of the networks by adjusting the content of disulfide bond and the degree of cross-linking, a broad glass transition temperature(T_g) was achieved, which enabled the network a versatile shape-memory effect, covering from dual-, triple-so far as to quadrupleshape memory effect. More importantly, the shape recovery and healing process can be realized simultaneously because of the highly matched actuating condition in this system.

摘要：Self-healing materials have attracted considerable attention because of their improved safety, lifetime, energy efficiency and environmental impact. Supramolecular interactions have been extensively considered in the field of self-healing materials due to their excellent reversibility and sensitive responsiveness to environmental stimuli. However,development of a polymeric material with good mechanical performance as well as self-healing capacity is very challenging. In this study, we report a robust self-healing polyurethane(PU) elastomer polypropylene glycol-2-amino-5-(2-hydroxyethyl)-6-methylpyrimidin-4-ol(PPG-mUPy) by integrating ureidopyrimidone(UPy) motifs with a PPG segment with a well-defined architecture and microphase *** balance the self-healing capacity and mechanical performance, a thermal-triggered switch of H-bonding is introduced. The quadruple H-bonded UPy dimeric moieties in the backbone induce phase separation to form a hard domain as well as enable further aggregation into microcrystals by virtue of the stacking interactions, which are stable in ambient temperature. This feature endows the PU with high mechanical strength. Meanwhile, a high healing efficiency can be realized, when the reversibility of the H-bond was unlocked from the stacking at higher temperature. An optimized sample PPG1000-mUPy50%with a good balance of mechanical performance(20.62 MPa of tensile strength) and healing efficiency(93% in tensile strength) was achieved. This strategy will provide a new idea for developing robust self-healing polymers.

摘要：本文基于所发展的多级压气机流动稳定性快速预测模型,首先应用激盘和半激盘模型对TA66低速实验台进行了失速起始点预测,并与实验结果对比,验证了模型的可靠性,讨论了不同径向模态与周向模态对预测结果的影响。进而用此模型预测了某型涡扇发动机的压气机稳定工作范围,为其稳定性设计提供依据,并逐级减少叶片排数量,观察系统稳定性变化,分析了多叶片排对系统稳定性带来的影响.

摘要：复杂地形条件下,基于深度强化学习的无人机编队航迹规划可以完成无人机编队的轨迹寻优,路径长度和环境适应性均优于传统启发式算法,但仍存在训练稳定性不足、规划实时性差等问题.面向领航者-跟随者模式的无人机集群,本文提出了一种基于SPER-TD3算法的无人机编队实时三维航迹规划方法.首先,将基于SumTree的优先经验回放机制融入TD3算法,设计了SPER-TD3算法,确定无人机编队的轨迹;然后,使用基于角度队形控制方法优化跟随者的飞行轨迹,并应用动态轨迹平滑算法优化转向角.为了加快SPER-TD3算法的训练收敛速度和稳定性,解决长时间依赖性问题,设计了结合LSTM、自注意力机制以及多重感知机的网络模型结构.在多种障碍物环境下进行了仿真实验,结果表明,所提方法在轨迹安全覆盖率、飞行路径平滑度、成功率、奖励大小等方面综合表现优于8种主流的深度强化学习算法,其重要性综合评估值比当前方法提升8.5%–72.9%不等,且训练稳定性最佳.

摘要：研究设计了CYCIAE-30回旋加速器主磁铁液压举升系统的油路结构和控制系统,并说明了它的工作原理。6年来的实际使用情况证明,该系统各项功能指标均达到了设计时的要求;油缸的实际最大工作行程为850mm,8个齐缝定位销回位重复定位精度为±0.01mm,两油缸同步精度可控制在5～10mm。该系统运行安全可靠,操作简便灵活。

摘要：基于PLAXIS 3D有限元软件,针对广州南沙某管廊基坑工程,系统分析了抽条加固、格栅状加固、裙边加固与满堂加固的面积置换率、加固宽度d、加固深度h等加固参数对基坑变形(围护结构侧移、地表沉降和坑底隆起)的影响,得到了加固参数的限值。结果表明:不同加固形式和加固参数对基坑变形的控制效果不同;4种加固方式均存在加固深度限值,抽条加固的加固深度限值为0.8H(H为基坑开挖深度),格栅状加固与满堂加固控制围护结构变形与地表沉降的加固深度限值为0.6H,控制坑底隆起的加固深度限值为0.8H,裙边加固深度大于0.6H时,控制效果显著;抽条加固、裙边加固不存在加固宽度限值,抽条加固宽度对基坑变形无影响,裙边加固增大加固宽度,对围护结构侧移与地表沉降的控制效果呈线性增加,加固宽度大于0.39B(B为基坑宽度)后对坑底隆起的控制效果显著,但裙边加固会使坑底中央隆起凸出,且加固宽度一般不超过0.3B,因此不建议在软土基坑工程中使用裙边加固;抽条加固、格栅状加固存在加固面积置换率限值,抽条加固控制基坑各项变形的面积置换率限值为66.67%,格栅状加固纵向面积置换率为64.29%时,中轴线存在加固区,隆起曲线呈“M”形,加固效果较好;加固面积相同时,加固效果从大到小依次为抽条加固、格栅状加固、裙边加固,加固体积相同时,加固效果从大到小依次为满堂加固、抽条加固、格栅状加固。

摘要：对低品位热驱动混合溶液除湿降温系统的性能进行了模拟研究.基于混合溶液传质关联式,对混合溶液除湿、再生过程进行了数学建模,比较了不同浓度溶液除湿器和再生器出口空气含湿量的实验值与模型计算值,其误差均在10%之内,从而验证了模型的可靠性.模拟分析了混合溶液浓度对系统性能的影响,模拟结果表明:在设计工况下,使用48%混合溶液的系统送风参数为19℃,9. 7 g/kg,氯化钙和氯化锂的混合溶液的除湿能力与氯化锂溶液相当,而混合溶液成本则大大降低,成本最高可降低65. 7%;随着混合溶液浓度的增加,系统热力性能提高且输配能耗降低,但也会导致送风温度及含湿量偏高,因此为了满足送风及节能需求,需要合理地选择氯化钙和氯化锂的混合溶液浓度.

摘要：基于脑电的脑机交互能帮助肢体运动障碍患者进行日常生活和康复训练,但是,由于脑电信号存在信噪比较低、个体差异性大等问题,导致脑电特征的提取与分类还需要进一步提高准确性和效率.因此,在减少脑电采集通道数目、增加分类数目的前提下,基于卷积神经网络对运动想象中的脑电信号进行分类.首先,基于已有方法进行探索实验,建立由 3层卷积层、3层池化层和 2层全连接层构成的卷积神经网络;然后针对想象左手、右手、脚的运动和静息态设计与开展了实验,获取了相关脑电数据;之后,利用脑电数据训练出基于卷积神经网络的分类模型,测试结果表明,该模型平均分类识别率达到了 82.81%,且高于已有的相关分类算法;最后,将已建立的分类模型应用于运动想象信号的在线分类,设计与开发了脑机交互应用原型系统,驱动人-机器人之间的实时交互,帮助用户利用运动想象控制仿人机器人的抬手、前进等运动状态.进一步的测试结果表明,机器人对用户控制命令的平均识别率达到了 80.31%,从而验证了所提方法可以对运动想象脑电数据进行较为精确的实时分类,可以促进脑机接口技术在人-机器人交互中的应用.

摘要：飞行器/发动机一体化是制约吸气式高超声速飞行器的核心和关键技术,本文针对高超声速飞发一体化构型开展了算力体系划分及算力参数敏感性研究。通过将一体化飞行器不同部件划分至气动和推进系统,研究了算力体系划分对气动/推进性能指标的影响,结果表明对于飞发一体化构型,在不同算力体系下表征的飞行器气动/推进性能可能存在较大差异,横向比较飞行器气动/推进性能必须在明确算力体系条件下进行。采用正交试验设计+方差分析的方法分析了飞行器算力对空域、速域、飞行姿态、气动热效应、真实气体效应5个因素的敏感性,结果表明壁面温度是影响飞行器轴向力计算的敏感参数,马赫数和攻角几乎影响所有气动指标。在研究范围内,雷诺数和气体比热比是飞行器气动性能的不敏感参数。

摘要：为了简化硬件实现的复杂度和降低存储量,提出一种采用码率预分配的JPEG2000码率控制算法,并给出相应的VLSI结构设计.原始图像经过小波变换和量化后,对EBCOT码块的有效比特平面进行独立熵估计,计算出所有码块的估计熵总和.依据每个码块的估计熵在所有码块的估计熵总和中所占的比例,指导分配每个码块的码率,EBCOT编码器根据分配到的码率实时截断码流和编码通道,减少了T1编码的时间.码块经过T1编码后直接打包输出,无需率失真斜率计算和率失真优化截取.实验结果表明,本算法很大程度上减少了编码计算量和存储量,易于硬件实现.