The multifunctional integration system(MFIS)is based on a common hardware platform that controls and regulates the system’s configurable parameters through software to meet dif-ferent operational requirements.Dwell s...The multifunctional integration system(MFIS)is based on a common hardware platform that controls and regulates the system’s configurable parameters through software to meet dif-ferent operational requirements.Dwell scheduling is a key for the system to realize multifunction and maximize the resource uti-lization.In this paper,an adaptive dwell scheduling optimization model for MFIS which considers the aperture partition and joint radar communication(JRC)waveform is established.To solve the formulated optimization problem,JRC scheduling condi-tions are proposed,including time overlapping condition,beam direction condition and aperture condition.Meanwhile,an effec-tive mechanism to dynamically occupy and release the aperture resource is introduced,where the time-pointer will slide to the earliest ending time of all currently scheduled tasks so that the occupied aperture resource can be released timely.Based on them,an adaptive dwell scheduling algorithm for MFIS with aperture partition and JRC waveform is put forward.Simulation results demonstrate that the proposed algorithm has better com-prehensive scheduling performance than up-to-date algorithms in all considered metrics.展开更多
关于汉语数量结构的研究成果颇丰,主要有两种观点,第一种观点以Cheng&Sybesma(1998/2005)的count classifiers和mass classifiers的区分为代表,第二种观点以Rothstein(2009/2010)、李旭平(2013)的classifiers of counting和classifi...关于汉语数量结构的研究成果颇丰,主要有两种观点,第一种观点以Cheng&Sybesma(1998/2005)的count classifiers和mass classifiers的区分为代表,第二种观点以Rothstein(2009/2010)、李旭平(2013)的classifiers of counting和classifiers of measure的区分为代表。本文介绍李旭平2013年出版的Numeral Classifiers in Chinese:The Syntax-Semantics Interface一书,介绍其汉语计数量词与计量量词的理论,并作简要评价。展开更多
针对特高压接入的输电网网架错综复杂、远景适应性差,甚至出现网架规划与开环分区决策相互独立导致资源浪费等问题,该文引入奇异值分解(singular value decomposition,SVD)和特征索引技术,提出了一种考虑电磁环网开环分区和500/220kV典...针对特高压接入的输电网网架错综复杂、远景适应性差,甚至出现网架规划与开环分区决策相互独立导致资源浪费等问题,该文引入奇异值分解(singular value decomposition,SVD)和特征索引技术,提出了一种考虑电磁环网开环分区和500/220kV典型供电结构的输电网多阶段规划方法。首先,提取典型供电结构的路径特征,制定了一种基于标签路径特征的典型供电结构索引策略;然后,构建了一种基于SVD的典型供电结构相似度指标,基于此,以全局典型供电结构相似程度最大为目标,建立了相似度识别规划模型;最后,提出了一种反馈调节机制及求解方法,统筹协调各阶段网架规划和开环分区之间的决策信息。与传统规划方法相比,该文方法不仅能够避免不同阶段开环分区决策导致的资源浪费,还可以减少复杂的规划评估指标计算,构建供电可靠性高、运行方式灵活、远景适应性强的网架结构。基于湖南某地区实际电网算例验证了该文方法的准确性和有效性。展开更多
针对利用海量数据构建分类模型时训练数据规模大、训练时间长且碳排放量大的问题,提出面向低能耗高性能的分类器两阶段数据选择方法TSDS(Two-Stage Data Selection)。首先,通过修正余弦相似度确定聚类中心,并将样本数据进行基于不相似...针对利用海量数据构建分类模型时训练数据规模大、训练时间长且碳排放量大的问题,提出面向低能耗高性能的分类器两阶段数据选择方法TSDS(Two-Stage Data Selection)。首先,通过修正余弦相似度确定聚类中心,并将样本数据进行基于不相似点的分裂层次聚类;其次,对聚类结果按数据分布自适应抽样以组成高质量的子样本集;最后,利用子样本集在分类模型上训练,在加速训练过程的同时提升模型精度。在Spambase、Bupa和Phoneme等6个数据集上构建支持向量机(SVM)和多层感知机(MLP)分类模型,验证TSDS的性能。实验结果表明在样本数据压缩比达到85.00%的情况下,TSDS能将分类模型准确率提升3~10个百分点,同时加速模型训练,使训练SVM分类器的能耗平均降低93.76%,训练MLP分类器的能耗平均降低75.41%。可见,TSDS在大数据场景的分类任务上既能缩短训练时间和减少能耗,又能提升分类器性能,从而助力实现“双碳”目标。展开更多
基金supported by the National Natural Science Foundation of China(6203100762371093).
文摘The multifunctional integration system(MFIS)is based on a common hardware platform that controls and regulates the system’s configurable parameters through software to meet dif-ferent operational requirements.Dwell scheduling is a key for the system to realize multifunction and maximize the resource uti-lization.In this paper,an adaptive dwell scheduling optimization model for MFIS which considers the aperture partition and joint radar communication(JRC)waveform is established.To solve the formulated optimization problem,JRC scheduling condi-tions are proposed,including time overlapping condition,beam direction condition and aperture condition.Meanwhile,an effec-tive mechanism to dynamically occupy and release the aperture resource is introduced,where the time-pointer will slide to the earliest ending time of all currently scheduled tasks so that the occupied aperture resource can be released timely.Based on them,an adaptive dwell scheduling algorithm for MFIS with aperture partition and JRC waveform is put forward.Simulation results demonstrate that the proposed algorithm has better com-prehensive scheduling performance than up-to-date algorithms in all considered metrics.
文摘关于汉语数量结构的研究成果颇丰,主要有两种观点,第一种观点以Cheng&Sybesma(1998/2005)的count classifiers和mass classifiers的区分为代表,第二种观点以Rothstein(2009/2010)、李旭平(2013)的classifiers of counting和classifiers of measure的区分为代表。本文介绍李旭平2013年出版的Numeral Classifiers in Chinese:The Syntax-Semantics Interface一书,介绍其汉语计数量词与计量量词的理论,并作简要评价。
文摘针对特高压接入的输电网网架错综复杂、远景适应性差,甚至出现网架规划与开环分区决策相互独立导致资源浪费等问题,该文引入奇异值分解(singular value decomposition,SVD)和特征索引技术,提出了一种考虑电磁环网开环分区和500/220kV典型供电结构的输电网多阶段规划方法。首先,提取典型供电结构的路径特征,制定了一种基于标签路径特征的典型供电结构索引策略;然后,构建了一种基于SVD的典型供电结构相似度指标,基于此,以全局典型供电结构相似程度最大为目标,建立了相似度识别规划模型;最后,提出了一种反馈调节机制及求解方法,统筹协调各阶段网架规划和开环分区之间的决策信息。与传统规划方法相比,该文方法不仅能够避免不同阶段开环分区决策导致的资源浪费,还可以减少复杂的规划评估指标计算,构建供电可靠性高、运行方式灵活、远景适应性强的网架结构。基于湖南某地区实际电网算例验证了该文方法的准确性和有效性。
文摘针对利用海量数据构建分类模型时训练数据规模大、训练时间长且碳排放量大的问题,提出面向低能耗高性能的分类器两阶段数据选择方法TSDS(Two-Stage Data Selection)。首先,通过修正余弦相似度确定聚类中心,并将样本数据进行基于不相似点的分裂层次聚类;其次,对聚类结果按数据分布自适应抽样以组成高质量的子样本集;最后,利用子样本集在分类模型上训练,在加速训练过程的同时提升模型精度。在Spambase、Bupa和Phoneme等6个数据集上构建支持向量机(SVM)和多层感知机(MLP)分类模型,验证TSDS的性能。实验结果表明在样本数据压缩比达到85.00%的情况下,TSDS能将分类模型准确率提升3~10个百分点,同时加速模型训练,使训练SVM分类器的能耗平均降低93.76%,训练MLP分类器的能耗平均降低75.41%。可见,TSDS在大数据场景的分类任务上既能缩短训练时间和减少能耗,又能提升分类器性能,从而助力实现“双碳”目标。
