卷积神经网络用于近红外光谱古筝面板木材分级

目前,我国乐器制作行业在古筝面板用木材等级的筛选上主要依赖于技师主观评判,但此法缺少科学理论的依据,效率低,客观性及出材率的提高等方面受到限制,无法满足乐器市场的大量需求。实现古筝面板用木材快速、智能化的分级工作是一个急需解决的课题。近红外光谱非常适用于测量含氢的有机物质。古筝面板木材主要化学成分的化学键均由含氢基团组成,不同等级板材的化学成分存在差异,这些差异反映在近红外光谱中,为判断木材等级提供了可能。同时卷积神经网络对非线性数据具有较强的特征提取能力,所以提出一种应用卷积神经网络模型对光谱数据进行分析的方法,进而判别木材的等级。应用了Savitzky Golay一阶、二阶微分两种预处理方法和核主成分分析、连续投影算法两种数据压缩方法,通过所设计的卷积神经网络模型以样本识别准确率和模型构建过程中的损失值作为判定指标选出最佳预处理和数据压缩方法。为了提高模型提取分析光谱数据的能力和避免过拟合现象,应用了多通道卷积核、批量归一化和early stopping策略,将通过两层卷积层提取的特征信息送入全连接层,从而充分提取剩余信息,通过Softmax函数获得板材的最终预测等级,从而确定了最终模型。最终Savitzky Golay一阶微分和核主成分分析为最佳数据处理方法,同时得出用于区分不同等级的古筝面板用木材的主要关键谱带,分别为1 163~1 243, 1 346~1 375和1 525~1 584 nm。将该模型应用于测试集样本,古筝面板用木材的等级识别准确率为95.5%。实验结果表明所提出的方法可以高效地处理光谱数据,有效识别区分不同等级的古筝面板用木材的关键特征,从而为广阔的乐器市场提供一定的技术支持。     

Salting‐out‐assisted liquid–liquid extraction with acetonitrile for the determination of trimetazidine in rat plasma using liquid chromatography–mass spectrometry

A high‐throughout bioanalytical method based on salting‐out‐assisted liquid/liquid extraction (SALLE) method with acetonitrile and mass spectrometry‐compatible salts followed by LC‐MS/MS analysis of trimetazidine in rat plasma is presented. It required only 50 μL of plasma and allows the use of minimal volumes of organic solvents. The seamless interface of SALLE and LC‐MS eliminated the drying‐down step and the extract was diluted and injected into an LC‐MS/MS system with a cycle time of 2.5 min/sample. The retention times of trimetazidine and IS were approximately 1.1 and 1.7 min, respectively. Calibration curves were linear over the concentration range of 0.1–100 ng/mL, which can be extended to 500 ng/mL by dilution. The intra‐ and inter‐batch precision, accuracy and the relative standard deviation were all <15%. This method was successfully applied to determine trimetazidine concentrations in rat plasma. Copyright © 2014 John Wiley & Sons, Ltd.     

双稳态结构中的1/2次谐波共振及其对隔振特性的影响

以典型的双稳态系统——屈曲梁结构为例,基于等效模型,结合解析、数值和实验手段,研究了双稳态结构中的1/2次谐波共振特性、演化过程、参数调节规律及其对隔振特性的影响.研究发现,当非线性刚度系数或激励幅值增加到一定程度时,系统会在一定带宽下产生显著的1/2次谐波共振;随着激励幅值增加,阻尼系统的1/2次谐波遵循“产生-增强-衰退-消失”的过程,该过程对峰值频率和峰值传递率有重要影响;适当提高非线性强度能有效改善双稳态结构隔振特性.针对双稳态屈曲梁结构开展的实验验证了1/2次谐波特性和隔振特性变化规律.     

100 MeV质子束流品质诊断及剂量测量准确性评估方法研究

质子辐射生物学效应是太空放射生物学和质子束放疗研究的重要基础，可为空间环境下人员的危险性评估以及质子治疗优化设计提供科学依据。依托加速器建立相应的生物样本辐照技术是开展此类研究的前提条件。中国原子能科学研究院最近建立的100 MeV强流质子回旋加速器提供的中能质子束流为目前国内能量最高，特别适合用于太空放射生物学和质子治疗相关研究。本研究中，利用在束和离线等多种手段建立了中能质子束流诊断和剂量测量方法，对加速器引出的100 MeV质子照射野大小、均匀性等束流品质以及剂量测量系统准确性进行了分析和评估。结果表明，对光子剂量响应好的LiF(Mg, Ti)热释光探测器，对90 MeV质子同样具有良好的剂量响应关系，可作为中能质子剂量准确性评估的手段之一。在5.0 cm×5.0 cm照射野范围内，加速器引出的100 MeV质子束流的均匀性好于90%，在线剂量测量系统准确性好于93%，束流品质和剂量测量条件基本满足辐射生物学的要求，可为质子辐射生物学效应研究的开展提供可靠保障。     

基于化学学科认知障碍突破的中考复习教学实践* ——以“二氧化碳的性质”为例

分析学生对二氧化碳的3个典型学科认知障碍,以“二氧化碳的捕捉”为情境,通过“尝试捕捉二氧化碳”“谁在捕捉二氧化碳”“捕捉背后的原理”等3个主要环节,实现对学科认知障碍的突破,展示了一种新的中考复习教学思路。     

基于深度学习的ELM实时识别研究

基于深度学习的方法，在HL-2A装置上开发出了一套边缘局域模(ELM)实时识别算法。算法使用5200次放电数据(约24.19万数据切片)进行学习，得到一个深度为22层的卷积神经网络。为衡量算法的识别能力，识别了HL-2A装置自2009年实现稳定ELMy H模放电以来所有历史数据(约26000次放电数据)，共识别出1665次H模放电，其中误识别35次，误报率为2.10%。在实际的1634次H模放电中，漏识别4次，漏识别率为0.24%。该误报率和漏报率可以满足ELM实时识别的精度要求。识别算法在实时控制环境下，对单个时间点的平均计算时间为0.46ms，可以满足实时控制的计算速度要求。     

Utilizing the Space‐Charge Region of the FeNi‐LDH/CoP p‐n Junction to Promote Performance in Oxygen Evolution Electrocatalysis

The modulation of electron density is an effective option for efficient alternative electrocatalysts. Here, p‐n junctions are constructed in 3D free‐standing FeNi‐LDH/CoP/carbon cloth (CC) electrode (LDH=layered double hydroxide). The positively charged FeNi‐LDH in the space‐charge region can significantly boost oxygen evolution reaction. Therefore, the j at 1.485 V (vs. RHE) of FeNi‐LDH/CoP/CC achieves ca. 10‐fold and ca. 100‐fold increases compared to those of FeNi‐LDH/CC and CoP/CC, respectively. Density functional theory calculation reveals OH^? has a stronger trend to adsorb on the surface of FeNi‐LDH side in the p‐n junction compared to individual FeNi‐LDH further verifying the synergistic effect in the p‐n junction. Additionally, it represents excellent activity toward water splitting. The utilization of heterojunctions would open up an entirely new possibility to purposefully regulate the electronic structure of active sites and promote their catalytic activities.     

Investigation of mononuclear,dinuclear, and trinuclear transition metal (II) complexes derived from an asymmetric Salamo‐based ligand possessing three different coordination modes

A new asymmetric Salamo‐based ligand H₂L was synthesized using 3‐tert‐butyl‐salicylaldehyde and 6‐methoxy‐2‐[O‐(1‐ethyloxyamide)]‐oxime‐1‐phenol. By adjusting the ratio of the ligand H₂L and Cu (II), Co (II), and Ni (II) ions, mononuclear, dinuclear, and trinuclear transition metal (II) complexes, [Cu(L)], [{Co(L)}₂], and [{Ni(L)(CH₃COO)(CH₃CH₂OH)}₂Ni] with the ligand H₂L possessing completely different coordination modes were obtained, respectively. The optical spectra of ligand H₂L and its Cu (II), Co (II) and Ni (II) complexes were investigated. The Cu (II) complex is a mononuclear structure, and the Cu (II) atom is tetracoordinated to form a planar quadrilateral structure. The Co (II) complex is dinuclear, and the two Co (II) atoms are pentacoordinated and have coordination geometries of distorted triangular bipyramid. The Ni (II) complex is a trinuclear structure, and the terminal and central Ni (II) atoms are all hexacoordinated, forming distorted octahedral geometries. Furthermore, optical properties including UV–Vis, IR, and fluorescence of the Cu (II), Co (II), and Ni (II) complexes were investigated. Finally, the antibacterial activities of the Cu (II), Co (II), and Ni (II) complexes were explored. According to the experimental results, the inhibitory effect was found to be enhanced with increasing concentrations of the Cu (II), Co (II), and Ni (II) complexes.     

砂岩方梁断裂过程区范围的实验确定方法

裂纹前端的断裂过程区是引起岩石非线性断裂及尺寸效应的主要原因。利用数字图像相关技术对砂岩开展了三点弯曲梁实验,获得观测区域高精度的全场位移和应变数据,根据断裂韧带区域水平位移和水平应变的分布特征,结合裂尖岩石颗粒变化的微观分析,提出采用裂纹尖端水平位移波动性和水平应变突变性所得到的波动系数和水平应变突变值,确定断裂过程区形状和临界尺寸的方法。结果表明:砂岩断裂过程区的形状为不规则的狭长带状区域,断裂过程区的临界长度为11~13mm,临界宽度为1.58~2.36mm。断裂过程区区域内形变在趋向裂尖时呈指数增加,但其单位区域内的形变增量呈波动状态。该方法能够更加准确判断岩石断裂过程区的范围,有助于分析岩石的非线性断裂特性。