关键词： driving pattern recognition   dual-motor battery electric vehicle   energy management strategy   hierarchical cluster method   principal component analysis  

摘要： To improve the economic performance of dual-motor battery electric vehicles, a novel driving pattern recognition-based energy management strategy (NDPREMS) is proposed in this paper. The NDPREMS firstly employs principal component analysis method to reduce the dimension of characteristic parameters of driving patterns and uses hierarchical cluster method for classifying driving patterns to construct a database of typical driving patterns, based on which a driving pattern recognizer is achieved using generalized regression neural network (GRNN) and the accuracy of this recognizer reaches 96.08%. In order to reasonably allocate the power between two motors, on the basis of rule-based energy management strategy (REMS), a dynamic programming-based energy management strategy (DPEMS) under typical driving patterns is formulated. By doing so, the logic thresholds of REMS are optimized, and thus, the NDPREMS is achieved. Comparison simulations of control effect concerning the REMS, DPEMS, and NDPREMS are performed under typical driving patterns. Results indicate that the proposed NDPREMS exhibits greater energy conservation compared with REMS, the economic improvement under urban driving pattern is the most obvious at 11.04%, the improvement under the comprehensive test driving pattern is 5.65%, and the performance of the NDPREMS is similar to that of DPEMS.

关键词： Comprehensive two-dimensional gas chromatography   GCxGC   Machine learning   Classification   Data mining  

摘要： Machine learning (ML) has been used previously to recognize particular patterns of constituent compounds. Here, ML is used with comprehensive chemical fingerprints that capture the distribution of all constituent compounds to flexibly perform various pattern recognition tasks. Such pattern recognition requires a sequence of chemical analysis, data analysis, and pattern analysis. Chemical analysis with comprehensive multidimensional chromatography is a maturing approach for highly effective separations of complex samples and so provides a solid foundation for undertaking comprehensive chemical fingerprinting. Data analysis with smart templates employs marker peaks and chemical logic for chromatographic alignment and peak-regions to delineate chromatographic windows in which analytes are quantified and matched consistently across chromatograms to create chemical profiles that serve as complete fingerprints. Pattern analysis uses ML techniques with the resulting fingerprints to recognize sample characteristics, e.g., for classification. Our experiments evaluated the effectiveness of seventeen different ML techniques for various classification problems with chemical fingerprints from a rich data set from 126 wine samples of different varieties, geographic regions, vintages, and wineries. Results of these experiments showed an accuracy range from 58% to 88% for different ML methods on the most difficult classification problems and 96% to 100% for different ML methods on the least difficult classification problems. Averaged over 14 classification problems, accuracy for the different methods ranged from 80% to 90%, with some relatively simple ML techniques among the top-performing methods. (C) 2019 Elsevier B.V. All rights reserved.

关键词： Varnish   Gene expression   Lacquer   Mass spectrometry   Gas chromatography  

摘要： A differential expression analysis technology developed for linear modeling of gene expression data was used in combination with thermally assisted hydrolysis and methylation gas chromatography/mass spectrometry (THM-GC/MS) to support the analysis of lacquers and varnishes on historical objects. Exudates from tropical trees, such as Manila copal, sandarac, South American copal, and Congo copal, which were frequently used in finishing layers on decorative objects up to the early 20th century, were compared through this approach. Highly discriminating features indicate biomarkers that can help to identify copals in resinous lacquers. The approach allows new, more systematic ways for finding biomarkers in the analysis of lacquered objects of art and varnishes.

关键词： 模式识别   信息处理   机器学习   微传感器   传感器信息融合   数学研究方法   MEMS   统计学习  

摘要： 本项目利用统计学习与机器学习等方法,针对MEMS传感器信息融合与处理技术和模式识别进行研究,属于数理科学在信息、仪器科学等学科中的交叉问题。针对微悬臂梁力学特性、多传感器的数据融合、MEMS水听器信号处理、超声信号缺陷分类识别以及优化算法等方面进行了较系统的研究工作,取得了一系列研究成果,丰富了该领域的理论成果和数学研究方法,对数理学科在信息科学等学科交叉问题中的应用具有重要的理论意义和实用价值。

关键词： 言语性幻听   影像遗传学   脑环路   精神障碍  

摘要： 言语性幻听(AVHs)特指在确无外界现实声音刺激的条件下,患者体验到有针对自己的"真实"声音存在,多数"声音"给患者带来精神痛苦,是高致残致死性精神症状之一;并且AVHs可出现于多种精神障碍患者中,极易导致疾病的误诊误治。因此,探寻AVHs的病理特征和精准治疗靶点已经成为当前亟需解决的问题。采用基于高通量测序的基因组学技术和基于脑影像的脑连接组学技术,探索分裂症、双相障碍、单纯抑郁、边缘型人格障碍和具有AVHs症状的健康者等6类患者AVHs症状的影像遗传学特征;并采用模式识别技术,基于上述特征,建立不同病种AVHs早期定性诊断和疗效预测体系将有助于言语性幻听的机制研究。笔者从内源性遗传特征-内表型特征(脑环路/脑网络)-表型特征(临床表现)3个角度揭示不同病种AVHs共性及特异性影像遗传学特征,以确定不同AVHs的精准治疗靶点,对建立早期诊疗预测模型、提高早期治愈率具有重要的科学意义。

关键词： 葛根   指纹图谱   HPLC   模式识别   质量评价  

摘要： 目的建立葛根药材高效液相色谱法( high performance liquid chromatography,HPLC)指纹图谱,为葛根质量评价提供依据。方法采用 Waters Acquity HPLC SunFire C 18 色谱柱( 250 mm×4. 6 mm,5 μm),以乙腈-0. 2%乙酸水溶液梯度洗脱,流速 0. 8 mL/min,柱温 30℃,检测波长为250 nm,建立葛根药材指纹图谱共有模式。将相似度评价与模式识别相结合,并构造质量评价函数对不同批次葛根药材进行质量评价。结果采用建立的葛根药材 HPLC 指纹图谱方法确定了 24 个共有峰,并对其中 5 个共有峰进行了指认,分别为 3'-羟基葛根素、葛根素、大豆苷、染料木苷、大豆苷元。12 批葛根药材相似度相似度均大于 0. 950,聚类分析与主成分分析结果一致,均将药材分为4 类。构造的质量评价函数表明聚为同类的药材函数值相近,并据此对所划分的 4 类葛根药材进行了质量评价,S5、S7、S11 质量较好,S2、S3、S4 质量较差,其余药材接近平均水平。结论葛根 HPLC指纹图谱重复性好,稳定可靠,可为葛根的质量控制和评价提供依据。

关键词： 陶瓷梭式窑   温度智能检测   模式识别   火焰图像   概率神经网络  

摘要： 陶瓷梭式窑是一种常用的间歇式陶瓷产品生产窑炉,其烧成带温度检测方法与陶瓷产品生产质量息息相关,因而研究其烧成带温度优化检测方法和算法具有重要意义。本文针对陶瓷梭式窑烧成带温度基于模式识别提出了一种优化检测方法和算法。首先对陶瓷梭式窑火焰图像进行预处理,利用L*a*b*颜色空间和灰度共生矩阵提取火焰图像的颜色特征和纹理特征构成特征向量,其次将特征向量输入到概率神经网络进行识别分类,最后利用温度信息数据与火焰图像得到的综合数据库来确定火焰图像对应的温度值从而识别陶瓷窑炉烧成带温度。实验结果表明所设计的火焰图像模式识别分类器具有较好的火焰图像识别率。因而,所提出的陶瓷梭式窑炉温度智能检测方法是可行和有效的。

关键词： 栀子   指纹图谱   去乙酰车叶草酸甲酯   京尼平龙胆双糖苷   栀子苷   西红花苷I   西红花苷II   化学模式识别  

摘要： 目的基于HPLC指纹图谱与多成分含量测定,并结合化学模式识别法,评价不同产地栀子的质量,优选最佳产地。方法采用RP-HPLC切换波长法,流动相为乙腈-0.1%磷酸水溶液进行梯度洗脱,检测波长为234、440nm,对10个不同产地的30批栀子药材进行指纹图谱及5种有效成分含量测定研究,应用聚类分析与主成分分析等化学模式识别方法对栀子质量进行评价。结果 10个不同产地30批栀子的指纹图谱相似度均在0.95以上,去乙酰车叶草酸甲酯、京尼平龙胆双糖苷、栀子苷、西红花苷I和西红花苷II质量分数分别为0.61~4.26、1.73~12.92、51.79~82.76、5.03~12.80、0.71~2.28mg/g;聚类分析可分为3类,第1类为产地湖南永州、湖南永安、江西丰城、江西湖口、安徽广德和福建福鼎;第2类为产地江西黎川、江苏句容和河南唐河;第3类为产地安徽六安。通过主成分分析,进一步评价不同产地间质量的差异,从而发现安徽六安、江西湖口和湖南永安等地质量较优。结论不同产地之间栀子存在一定质量差异,同一产地不同批次间栀子质量较稳定。通过指纹图谱与含量测定相结合,应用化学模式识别方法可全面综合评价栀子质量。此方法的建立为栀子药材的质量控制及评价提供了有效的参考依据。