基于循环谱和深度神经网络的调制识别算法

针对通信调制信号的循环谱存在数据量大、采用神经网络训练耗时等问题,提出了一种联合自编码器(Auto-Encoder,AE)和卷积神经网络(Convolutional Neural Network,CNN)同时完成数据降维和调制识别的算法——AE-CNN。该算法以循环频率轴的投影作为输入,采用AE实现对数据压缩,并将AE隐层输出至CNN,通过联合训练实现对6种不同信号的调制识别。同时分析了基于CNN、主成分分析(Principal Component Analysis,PCA)和AE三种方法的识别性能。实验结果表明,与基于PCA和AE的算法相比,AE-CNN的识别性能分别提高了约1.6%和2.1%;相对于基于CNN的方法能够大幅减少训练时间,并在识别性能上提升约1.2%。

数据驱动的双通道CNN-LSTM调制分类算法

为同时利用卷积神经网络(Convolutional Neural Network, CNN)的空间特征提取能力和长短时记忆(Long Short-Term Memory, LSTM)网络的时序特征提取能力,提出了一种由双通道一维CNN与LSTM相互串联的调制分类算法。算法采用数据驱动的方式,直接将信号送入至2路CNN提取其在不同维度的空间特征信息;把2个通道的特征融合信息输入至LSTM学习其时序上的特征;与全连接网络连接实现对5种目标信号的调制分类。实验结果表明,CNN与LSTM相互串联能够学习到更加丰富的特征信息,更有利于分类;与传统方法相比,提出的方法无需人工提取信号特征,减少了预处理步骤并有效提升了识别性能。

基于星座图和密集连接网络的QAM信号识别

利用深度神经网络对图像数据的显著学习能力,提出了一种基于通信信号星座图和密集连接网络(DenseNet)的正交幅度调制(Quadrature Amplitude Modulation,QAM)信号调制分类算法。算法首先对接收信号进行预处理,获取训练所需的星座图数据集,然后采用DenseNet对其进行训练学习,进而实现调制分类。同时在DenseNet网络中引入通道注意力机制,进一步增强特征的学习能力,提升分类性能。对不同信号长度、不同神经网络以及存在频偏和相偏估计误差等场景进行了多组实验。仿真结果表明:DenseNet相比卷积神经网络(Convolutional Neural Network,CNN)和残差网络ResNet-20能够有效提升分类准确率;相比基于累积量、累积分布及原始IQ(In-phase and Quadrature)数据等现有方法也均取得了较优的识别性能。

320排CT全肝脾联合灌注对肝脏早期弥漫性病变的临床应用价值

目的探讨320排CT全肝脾联合灌注成像应用于肝脏早期弥漫性病变诊断的可行性及优越性。方法 23例正常肝脏志愿者(正常组)、23例肝纤维化患者(病例组)均用320排CT行全肝脾联合动态容积扫描,所获得图像利用灌注软件获取兴趣区的时间-密度曲线及灌注参数,并测量脾脏大小,然后应用合适的统计学方法对正常组及病例组各数据进行分析。结果病例组门静脉(PV)、肝脏及肝纤维化F4级脾脏的曲线峰值明显滞后,波峰宽大,且达峰后下降缓慢,形成较长时间平台期,峰值不明显。正常肝脏各肝叶肝动脉灌注量(AF)、门静脉灌注量(PF)、肝灌注指数(PI)差异均无统计学意义(P>0.05);病例组与正常组肝脏各叶PF差异有统计学意义(P<0.01),AF及PI差异无统计学意义(P>0.05);病例组与正常组的脾脏灌注参数差异有统计学意义(P<0.05);病例组中肝纤维化F2级与F3级脾脏AF及肝纤维化F2~F4级脾脏厚度的比较差异无统计学意义(P>0.05),余肝纤维化分组间的脾脏AF差异均有统计学意义(P<0.05)。结论 320排CT动态容积全肝脾灌注成像从整体上评价肝脾微循环变化,对肝脏早期弥漫性病变的灌注分析有显著优势,从而为肝脏早期弥漫性病变的诊断及分级提供了更多依据。

Deep-blue organic light-emitting diodes based on a doublet d-f transition cerium(Ⅲ) complex with 100% exciton utilization efficiency

Compared to red and green organic light-emitting diodes(OLEDs),blue OLEDs are still the bottleneck due to the lack of efficient emitters with simultaneous high exciton utilization efficiency(EUE)and short excited-state lifetime.Different from the fluorescence,phosphorescence,thermally activated delayed fluorescence(TADF),and organic radical materials traditionally used in OLEDs,we demonstrate herein a new type of emitter,cerium(Ⅲ)complex Ce-1 with spin-allowed and parity-allowed d-f transition of the centre Ce^(3+) ion.The compound exhibits a high EUE up to 100% in OLEDs and a short excited-state lifetime of 42 ns,which is considerably faster than that achieved in efficient phosphorescence and TADF emitters.The optimized OLEDs show an average maximum external quantum efficiency(EQE)of 12.4% and Commission Internationale de L’Eclairage(CIE)coordinates of(0.146,0.078).

Room-temperature phosphorescence from purely organic materials

Room-temperature phosphorescence（RTP） materials have attracted great attention due to their involvement of excited triplet states and comparatively long decay lifetimes.In this short review,recent progress on enhancement of RTP from purely organic materials is summarized.According to the mechanism of phosphorescence emission,two principles are discussed to construct efficient RTP materials：one is promoting intersystem crossing（ISC） efficiency by using aromatic carbonyl,heavyatom,or/and heterocycle/heteroatom containing compounds;the other is suppressing intramolecular motion and intermolecular collision which can quench excited triplet states,including embedding phosphors into polymers and packing them tightly in crystals.With aforementioned strategies,RTP from purely organic materials was achieved both in fluid and rigid media.