关键词： memory   multibit   two-dimensional(2D)   MoSSe   Co-modulation  

摘要： The explosively developed era of big-data compels the increasing demand of nonvolatile memory with high efficiency and excellent storage ***,we fabricated a high-speed photoelectric multilevel memory device for neuromorphic *** novel two-dimensional(2D)MoSSe with a unique Janus structure was employed as the channel,and the stack of Al_(2)O_(3)/black phosphorus quantum dots(BPQDs)/Al_(2)O_(3)was adopted as the *** storage performance of the resulting memory could be verified by the endurance and retention tests,in which the device could remain stable states of programming and erasing even after 1,000 cycles and 1,000 *** multibit storage could be realized through both different voltage amplitudes and pulse numbers,which could achieve 6 bits(64 distinguishable levels)under pulse width of 50 ***,our memory device also could realize the simulations of synapses in human brain with optical and electric modulations synergistically,such as excitatory post-synaptic current(EPSC),long-term potentiation/depression(LTP/LTD),and spike-timing-dependent plasticity(STDP).Neuromorphic computing was successfully achieved through a high recognition of handwritten digits up to 92.5%after 103 *** research is a promising avenue for the future development of efficient memory and artificial neural network systems.

关键词： 不安全行为   化学实验室   模式识别   三维卷积网络  

摘要： 针对因人的不安全行为导致化学实验室安全事故的预防问题,提出了基于三维卷积网络(Convolutional 3D,C3D)的化学实验室人员不安全行为模式识别方法。定义了5种典型的人员不安全行为模式,以某高校化学实验室作为研究区域,构建了人员不安全行为模式数据集,建立了基于C3D的人员不安全行为模式识别模型,并对模型在不同应用场景下的性能进行了评估。结果表明:在特定人员于特定实验场景中出现上述5种不安全行为时,该模型在测试集上得到的F1指数平均值均能达到97%以上;在非特定人员与非特定场景中出现不安全行为时,模型能够有效识别部分不安全行为。该研究成果预期可以为化学实验室人员不安全行为的预测、预警、预防提供技术支持。

关键词： antifungals   stewardship   prescriptions   children   amphotericin B   azoles  

摘要： Objective: The need for pediatric antifungal stewardship programs has been driven by an increasing consumption of antifungals for prophylactic and empirical use. Drivers and rational of antifungal prescribing need to be identified to optimize prescription behaviors. Methods: A prospective modified weekly Point Prevalence Survey capturing antifungal prescriptions for children (> 90 days to < 18 years of age) in 12 centers in England during 26 consecutive weeks was performed. Demographic, diagnostic and treatment information was collected for each patient. Data were entered into an online REDCap database. Results: One thousand two hundred fifty-eight prescriptions were included for 656 pediatric patients, 44.9% were girls, with a median age of 6.4 years (interquartile range, 2.5-11.3). Most common underlying condition was malignancy (55.5%). Four hundred nineteen (63.9%) received antifungals for prophylaxis, and 237 (36.1%) for treatment. Among patients receiving antifungal prophylaxis, 40.2% did not belong to a high-risk group. In those receiving antifungal treatment, 45.9%, 29.4%, 5.1% and 19.6% had a diagnosis of suspected, possible, probable of proven invasive fungal disease (IFD), respectively. Proven IFD was diagnosed in 78 patients, 84.6% (n = 66) suffered from invasive candidiasis and 15.4% (n = 12) from an invasive mold infection. Liposomal amphotericin B was the most commonly prescribed antifungal for both prophylaxis (36.6%) and empiric and preemptive treatment (47.9%). Throughout the duration of the study, 72 (11.0%) patients received combination antifungal therapy. Conclusions: Antifungal use in pediatric patients is dominated by liposomal amphotericin B and often without evidence for the presence of IFD. A significant proportion of prophylactic and empiric antifungal use was seen in pediatric patients not at high-risk for IFD.

关键词： Authenticity   FT-IR spectral fingerprint   Classification   Linear discriminant analysis   Tea   Geographical origin  

摘要： The potential of FT-IR spectra was examined to classify tea samples based on the geographical origins. Principal component analysis (PCA), principal component analysis-linear discriminant analysis (PCA-LDA) and partial least square-discriminant analysis (PLS-DA) were investigated in order to achieve discrimination of tea samples. Several spectral pre-processing methods, such as mean centering (MC), auto-scaling, multiplicative scatter correction (MSC), standard normal variate (SNV) and their combinations, were employed to improve the quality of the spectra. The results showed that the tea samples from five geographical regions can be identified based on using FT-IR spectral fingerprints. The results demonstrated that FT-IR spectral fingerprinting combined with pattern recognition methods can be employed as an effective and feasible method for classification of Iranian tea based on their geographical origins.

关键词： Condition monitoring   Multisource pattern recognition   Milling processing   Tool wear   Pattern transfer path  

摘要： In the milling process of metallic parts, appropriate tool conditions are essential to reducing processing faults and ensuring manufacturing quality. However, the existing condition monitoring methods are usually limited by recognizing intermediate abnormal states during milling processing, which is inefficient and impractical for real practical applications. Therefore, this paper proposes a tool condition monitoring (TCM) method in the milling process based on multisource pattern recognition and state transfer paths. First, the improved K-means clustering method is used to generate multiple patterns of tool wear. Second, a multisource pattern recognition model framework is developed, and multiple observation windows and the pattern transfer path are considered in the multisource pattern recognition model. Finally, PHM2010 datasets are used to verify the feasibility of the proposed method, and the results demonstrate the applicability of the proposed method in practice for tool condition monitoring.

关键词： Electric vehicle range   estimation   driving pattern recognition   clustering   energy consumption  

摘要： Limited driving range and availability of charging infrastructures are still among the main barriers of adoption of electric vehicles (EVs) in the market. Combination of those limiting factors causes 'range anxiety' in EV users. While different EV battery technologies and charging infrastructures are under development, one short-term solution to reduce EV users' range anxiety is to provide the EV user with an accurate range estimation. In this study, an EV range estimation technique is proposed that recognises the current driving pattern and then classifies it into one of the predefined clusters (driving modes). The future energy consumption per kilometre is then tuned according to the average energy consumption of each cluster. Having an updated energy consumption rate, the EV range is calculated based on the battery state-of-charge. Different features are considered for driving pattern clustering where 'average speed' and 'average power' were identified as the best choices for this application. The effectiveness of the proposed EV range estimator is validated using real driving data that gives an average error of 9% in EV energy consumption estimation ahead.

关键词： Amyloid beta   Poly   Machine learning   Pattern recognition   Conjugated polymers   Graphite oxide  

摘要： Five fluorescent positively charged poly(para-aryleneethynylene) (P1-P5) were designed to construct electrostatic complexes C1-C5 with negatively charged graphene oxide (GO). The fluorescence of conjugated polymers was quenched by the quencher GO. Three electrostatic complexes were enough to distinguish between 12 proteins with 100% accuracy. Furthermore, using these sensor arrays, we could identify the levels of A beta 40 and A beta 42 aggregates (monomers, oligomers, and fibrils) via employing machine learning algorithms, making it an attractive strategy for early diagnosis of Alzheimer's disease.

关键词： AlexNet模型   断层识别   模式识别   批量归一化   平衡交叉熵损失  

摘要： 从地震数据中识别断层在地震资料解释中至关重要,但随着勘探规模的扩大,传统的人工解释断层已满足不了实际生产需要。如何研究出一种能满足断层识别高精度需求并能提升运算速度的方法是急需解决的问题。为此,基于改进的AlexNet模型,把自动识别断层的方法看作图像识别二分类问题。首先将批量归一化代替局部响应归一化,加快模型收敛;其次引入平衡交叉熵损失,解决在地震数据中断层与非断层高度不平衡问题,使模型朝着正确的方向收敛;最后用卷积层代替全连接层,极大缩减了训练参数,加快了训练速度。训练的模型对理论数据和实际数据预测结果表明,改进的AlexNet模型充分学习了断层特征,具有可以从地震数据中识别断层的能力。

关键词： 三子散   高效液相色谱法   指纹图谱   相似度评价   化学模式识别   聚类分析   主成分分析   正交偏最小二乘法-判别分析   质量控制  

摘要： 目的建立蒙药三子散高效液相色谱(HPLC)指纹图谱,通过化学模式识别方法综合评价其内在质量。方法采用HPLC法,以京尼平苷色谱峰为参照峰,采用《中药色谱指纹图谱相似度评价系统(2012版)》绘制15批三子散的HPLC指纹图谱并进行相似度评价,标定并指认共有峰。结合聚类分析、主成分分析和正交偏最小二乘法-判别分析对15批三子散的指纹图谱进行质量评价,筛选影响其质量的差异性标志物。结果 15批三子散中共有29个共有峰,相似度均不低于0.952,表明15批三子散的化学组成一致性较好;共指认了诃子次酸、没食子酸、石榴皮鞣素、安石榴苷A、安石榴苷B、jasminoside B、咖啡酸、柯里拉京、京尼平苷、诃子鞣酸、1,2,3,4,6-O-没食子酰葡萄糖、诃子林鞣酸、鞣花酸13个共有峰。聚类分析与主成分分析均可将15批三子散分为4类,且分类结果一致,表明15批三子散的质量存在一定差异;结合正交偏最小二乘法-判别分析筛选出导致批次间质量差异的14个差异性标志物(诃子次酸、没食子酸、鞣花酸等)。结论建立的HPLC指纹图谱分析方法简便、稳定,结合化学模式识别分析可用于三子散的质量控制。

关键词： GIS设备   局部放电   Gabor变换   聚类分析   模式识别   机器学习  

摘要： 气体绝缘组合开关设备(Gas Insulated Switchgear,GIS)是电力系统中的关键设备,通过局部放电信号获取GIS设备内部绝缘缺陷的放电类型对于故障诊断和预警至关重要。通过试验提出了一种GIS设备不同放电缺陷类型的识别方法,首先搭建了GIS内典型缺陷模型并使用特高频传感器获取局部放电信号,随后对GIS局部放电的三维PRPS图谱进行Gabor变换,并对分解图提取纹理和形状特征量。使用层次聚类法对提取的特征量进行聚类分析,聚类结果表明特征量与缺陷的放电特征具有较好的关联程度,验证了根据PRPS图谱经Gabor变换后提取特征量对典型缺陷进行识别的可行性。基于此,对不同典型放电缺陷使用不同种类的机器学习算法对放电类型进行识别。结果表明:多种机器学习算法对3种典型放电类型的识别准确率较高,PRPS图谱经Gabor变换后提取出的特征量能够较好地反映放电特征,具有较高的区分度,该诊断方法能对GIS故障预警提供可靠的参考依据。